A method of data transfer, a method of controlling use of data and a cryptographic device

A method of data transfer, a method of controlling use of data and a cryptographic device

This is a divisional application of Canadian Patent Application Serial No. 3,013,687 filed on February 3, 2017.

Field

The present invention relates to methods of data transfer, methods of controlling the use of data, and cryptographic devices.

It should be understood that the expression “the invention” and the like used herein may refer to subject matter claimed in either the parent or the divisional applications.

Background

The internet has resulted in many organisations using computer implemented services hosted by a service provider, which previously they would have been required to host themselves. Service providers are able to provide their computer implemented services to a large number of organisations (tenants). An example of this is cloud service providers, CSPs, who offer products such as software as a service (SaaS) or storage on demand. Use of a service provider hosted computer implemented service allows the tenants to reduce the administrative overhead of hosting such services themselves.

An area in which service providers have struggled to offer a multi-tenanted computer implemented service is in the cryptographic infrastructure sector, and specifically in hardware security modules, HSMs.

A conventional HSM hosted by a service provider uses a single-tenancy solution, i.e. deploying a dedicated appliance per tenant, to enable the secure cryptographic processing of data and storage of cryptographic keys on behalf of the tenant. The service provider manages the environmental settings of the cryptographic appliance, such as the IP address,

Date Reçue/Date Received 2021-06-25

1a whilst the tenant manages the cryptographic infrastructure remotely in a conventional manner. The service provider thus has to provide an appliance per tenant, whilst the tenant has to manage the maintenance and administration of the cryptographic appliance much as they would have previously. Such a system is inefficient and often results in underutilisation of cryptographic resources.

In addition, such a system is may be vulnerable. The service provider often has the keys to the kingdom, that is, the ability to export the raw key material stored within the

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

HSM. The permissions granted to the service provider can thus undermine the security of a hosted tenant key. This results in a number of issues for both the tenant and the service provider, including possible rogue use of a tenant's cryptographic key by a service provider, exposure of a tenant's key to a security agency that has jurisdiction over the service provider but not the tenant, and use of a tenanfs key by another tenant.

Statements of Invention

In a first aspect of the present invention, there is provided a method of data transfer between a first security context in a tenant system and a second security context in a service provider system, the method comprising:

generating an access control list corresponding to the data in the first security context, wherein the access control list specifies that a valid use credential must be presented in order to grant a first type of use of the data;

generating a first cryptographic key pair and a first cryptographic certificate in the second security context, the first cryptographic key pair comprising a first public key, Kblob pub. and a first private key, K_Blob pdv and the first cryptographic certificate comprising information from which the origin of the first public key Kblob pub can be validated;

sending the first public key Kblob pub and the first cryptographic certificate to the first security context;

validating the first cryptographic certificate in the first security context;

if the first cryptographic certificate is valid, encrypting the data and the coresponding access control list with the first public key Kblob pub in the first security context;

sending the encrypted data and corresponding access control list, and information from which the origin of the data can be validated, to the second security context

In an embodiment, the use credential is a use certificate.

In an embodiment, the data comprises a cryptographic key, Kt_enant. The method may further comprise the step of generating the cryptographic key, Kt_enant in the first security context The first type of use may be one or more cryptographic operations.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

In an embodiment, the method further comprises establishing at the tenant system that the second security context is trusted, before sending the encrypted data.

Establishing trust may comprise validating at the first security context that the second security context is manufactured by a trusted manufacturer, that the configuration of the second security context meets the security requirements of the tenant and that the second security context is configured to enforce the policies contained in the ACL.

In an embodiment, it is validated that the configuration of the second security context meets the security requirements of the tenant immediately before transfer of the encrypted data to the second security context.

Establishing trust may further comprise validating that the state of the second security context meets the security requirements of the manufacturer, for example, validating that the software and hardware is impervious to attack by the service provider.

A manufacturer private key may be used to validate that the second security context is manufactured by a trusted manufacturer, and to validate a second identity public key of the second security context.

In an embodiment, the second security context stores a second identity private key, Kzid prhz, and the method further comprises sending the second identity public key, K_21Dpub, and a second identity certificate from the second security context to the first security context, wherein the second identity public key, K₂id pub and the second identity private key, K_2C are θ cryptographic key pair and the second identity certificate comprises information identifying Ka_{D pub} and is cryptographically signed by the manufacturer private key ΚύπρονIn an embodiment, the second identity certificate further comprises information from which the state of the second security context can be validated.

In an embodiment, the method further comprises establishing at the tenant system that a reference time source is trusted. In an embodiment, the method further comprises

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 establishing at the service provider system that a reference time source is trusted. Establishing that a reference time source is trusted comprises validating that the reference time source is manufactured by a trusted manufacturer and that the state and configuration of the reference time source meets the security requirements.

In an embodiment, the method further comprises:

generating information relating to the current configuration of the second security context;

cryptographically signing the information with the second identity private key, KziD privl sending the signed information from the second security context to the first security context.

In an embodiment, the first cryptographic certificate comprises the information relating to the current configuration of the second security context and is signed with the second identity private key, K₂,_D

In an embodiment, the method further comprises:

generating a second cryptographic key pair and a second cryptographic certificate in the first security context, the second cryptographic key pair comprising a second public key, K_tenant__sjgnpub, and a second private key, K_tenant__Sig_nprlv and the second cryptographic certificate comprising information from which the origin of the second public key K_tenant-sign_pub can be validated;

sending the second public key K_tenant__signpub and the second cryptographic certificate to the second security context.

In an embodiment, the first security context stores a first identity private key, Κ_{10 priv}, and the method further comprises:

sending a first identity public key, K_{1C pub}, and a first identity certificate from the first security context to the second security context, wherein the first identity public key, Ki io pub and the first identity private key, Kuo priv are a cryptographic key pair and the first identity certificate comprises information identifying K_1)D pu_b and is cryptographically signed by a manufacturer private key K_man

Date Reçue/Date Received 2021-06-25

In an embodiment, the second cryptographic certificate comprises information from which the origin of the second public key K_tenant__sjgnpub can be identified; and the method farther comprises:

cryptographically signing the second cryptographic certificate with the first identity private key, Ki ID priv!

verifying the second cryptographic certificate using the first identity public key, Ki id pub at the second security context.

In some embodiments, the method farther comprises:

generating a second cryptographic key pair and a second cryptographic certificate in the first security context, the second cryptographic key pair comprising a second public key, Ktenant-_Sign_pub. and a second private key, K_tenant__signprtv and the second cryptographic certificate comprising information from which the origin of the second public key Ktenant-Signpub ^{can 00} identified;

cryptographically signing the second cryptographic certificate with the first identity private key, Κι iDpdv;

sending the second public key K_tenant__signpub and the signed second cryptographic certificate to the second security context;

verifying the second cryptographic certificate using the first identity public key, Ki id pubThe information from which the origin of the cryptographic key Ktenant can be validated may comprise the encrypted cryptographic key K_tenant and corresponding access control list Signed With Ktenant-signp_r|_v'

The step of sending the encrypted cryptographic key Ktenant and corresponding access control list, and information from which the origin of the cryptographic key Ktenant can be validated, to the second security context may comprise:

Date Reçue/Date Received 2021-06-25

5a cryptographically signing the encrypted cryptographic key Ktenant and corresponding access control list with K_tenant__signprlv;

sending the encrypted cryptographic key K_tenant and corresponding access control list, the the signature of the encrypted cryptographic key Ktenant and corresponding access control list, and the hash of Ktenant-sign_pub to the second security context.

In some embodiments, the access control list specifies that the cryptographic key, Ktenant can only be encrypted for storage outside of the second security context by a key which cannot leave the second security context.

In an embodiment, the first cryptographic certificate validates that Kblobpw is ephemeral and that Kblobptîv cannot leave the second security context. In an embodiment, the first cryptographic certificate validates that the asymmetric key pair was generated in the second security context. The first private key, KgLospnv is stored in the second security context.

In an embodiment, the information from which the origin of the first public key KeLoepub can be validated is a signed hash of the first public key, Ketoepub. The first cryptographic certificate comprises a hash of the first public key, Kblobpuô and is signed with the private half of the identity key of the second security context, K₂id pnv- The step of validating the first cryptographic certificate, Cblob, comprises verifying the signature using the public half of the identity key of the second security context, K₂id pubDate Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

In an embodiment, the second security context is secure from the rest of the service provider system.

In an embodiment, the access control (ACL) list specifies that the use credential must comprise information from which the origin of the use credential can be validated. The access control list may specify that the use credential is a use certificate which must be signed by the second private key K_tenant-sign_prlv in order to grant the first type of use of the cryptographic key, Kt_enart

The ACL specifies that the use credential must comprise information from which the expiry of the use credential can be determined and must be unexpired in order to be valid.

The ACL may specify that the tenant cryptographic key Ktenam can only be stored in non-volatile memory being resistant to tamper by a third party.

In an embodiment, the ACL specifies that the tenant cryptographic key K_tenant can only be stored inside the second security context. In an alternative embodiment, the ACL contains a restriction that the tenant cryptographic key Ktenam can only be stored on the condition that it is encrypted for storage by a key which cannot leave the second security context.

In an embodiment, the method further comprises;

validating, at the second security context, the origin of the cryptographic key ^tenant decrypting the encrypted cryptographic key Ktenam and corresponding access control list with the first private key Κβιοβρ^δΛ the second security context

In an embodiment, the method further comprises;

re-encrypting the cryptographic key Kt_enam with a further cryptographic key at the second security context, wherein the further cryptographic key cannot leave the second security context;

storing the re-encrypted cryptographic key Kt_enam, corresponding access control list, and information from which the origin of the cryptographic key K^nant can be validated.

Date Reçue/Date Received 2021-06-25

In a further aspect ofthe present invention, there is provided a carrier medium comprising computer readable code configured to cause a computer to perform any of the methods described.

In a further aspect ofthe present invention, there is provided A method of controlling use of data, the data being stored in a service provider system in a manner which is accessible to a trusted second security context in the service provider system but secure from the rest of the service provider system, wherein an access control list specifying that a valid use credential must be presented in order to grant a first type of use ofthe data is stored with the data, wherein the data comprises a cryptographic key tenant, and wherein the access control list specifies that the cryptographic key, Ktenant can only be encrypted for storage outside ofthe second security context by a key which cannot leave the second security context, the method comprising:

generating a use credential in a first security context, wherein the use credential comprises:

information from which the data corresponding to the use credential can be identified;

information from which the expiry ofthe use credential can be determined;

issuing the use credential and information from which the origin ofthe use credential can be validated;

validating the use credential with respect to the access control list, and validating that the use credential has not expired at the second security context;

granting the first type of use ofthe data, in the second security context, on the condition that the use credential is valid and not expired.

In an embodiment, the use credential is a use certificate.

In an embodiment, the data comprises a cryptographic key, Ktenant. The first type of use may be one or more cryptographic operations.

In some embodiments, the use credential and information from which the origin ofthe use credential can be validated is issued to the service provider, the method further comprising:

Date Reçue/Date Received 2022-12-22

7a providing the use credential, the information from which the origin ofthe use credential can be validated, and a request to perform an operation with the key Ktenant, the operation being a first type of use, to the second security context;

performing the operation at the second security context, on the condition that the use credential is valid and not expired.

In an embodiment, the information from which the expiry ofthe use credential can be determined comprises:

an expiry time;

information identifying a reference time source.

Date Reçue/Date Received 2022-12-22

WO 2017/134445 PCT/GB2017/050264

In an embodiment, the method further comprises establishing at the tenant system that a reference time source is trusted. In an embodiment, the method further comprises establishing at the service provider system that a reference time source is trusted. Establishing that a reference time source is trusted comprises validating that the reference time source is manufactured by a trusted manufacturer and that the state and configuration of the reference time source meets the security requirements.

In an embodiment, the method further comprises providing the public half of an identity cryptographic key pair of the time source to the first security context, together with information validating the origin of the identity cryptographic key pair.

In an embodiment, the method further comprises providing the public half of an identity cryptographic key pair of the time source to the second security context, together with information validating the origin of the identity cryptographic key pair.

In an embodiment, generating a use credential in a first security context comprises: selecting a reference time source;

requesting the current time stamp from the reference time source calculating the expiry time based on the time stamp.

In an embodiment, the method comprises:

sending a message comprising the current time stamp from the reference time source to the tenant system, together with information from which the origin of the message can be validated; and validating the origin of the message.

The message may further comprise information relating to the current configuration of the reference time source.

The information validating the origin of the message may be the signed message, signed with the private half of the identity cryptographic key pair of the time source.

In an embodiment, information relating to a start time is included in the use credential.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

The information from which the data corresponding to the use credential can be identified may be a hash of Kt_enanl.

In an embodiment, the use credential is a use certificate and the method further comprises:

cryptographically signing the use certificate with a private key K_tenant__signpi._iv in the first security context, wherein the information from which the origin of the use certificate can be validated is the signature and wherein the corresponding public key ^Ktenant-sign_pub is integrity protect and accessible to the second security context;

verifying the use certificate using the public key K_tenant__signpub at the second security context.

In an embodiment, the method further comprises:

providing the use credential, the information from which the origin of the use credential can be validated, and a request to perform an operation with the key K_tenant. the operation being a first type of use, to the second security context;

performing the operation at the second security context, on the condition that the use credential is valid and not expired.

In an embodiment, validating that the use credential has not expired at the second security context comprises:

requesting the current time stamp from the reference time source;

sending a message comprising the current time stamp from the reference time source to the second security context, together with information from which the origin of the message can be validated; and validating the origin of the message.

The information validating the origin of the message may be the signed message, signed with the private half of the identity cryptographic key pair of the time source.

Validating that the use credential has not expired at the second security context may further comprise:

comparing the time stamp to the expiry time.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

In a further aspect of the present invention, there is provided a carrier medium comprising computer readable code configured to cause a computer to perform any of the methods described.

In a further aspect of the present invention, there is provided a cryptographic device comprising a first security context, the first security context comprising:

a first transceiver configured to receive a first public key Kblob put> and a first cryptographic certificate, comprising information from which the origin of the first public key Kblob pub can be validated, from a second security context;

a first processor configured to perform cryptographic operations, the first processor being configured to:

generate an access control list corresponding to data to be transferred, wherein the access control list specifies that a valid use credential must be presented in order to grant a first type of use of the data;

validate that the first cryptographic key pair originated from the second security context;

encrypt the data and the corresponding access control list with the first public key Kblob pub!

wherein the first transceiver is configured to send the encrypted data and corresponding access control list, and information from which the origin of the data can be validated, to the second security context.

In an embodiment, the use credential is a use certificate.

In an embodiment, the data comprises a cryptographic key, K^nantIn an embodiment, the device further comprises:

a first device memory, storing a first identity private key, Kn_Dpn_V;

wherein the first transceiver is further configured to:

send a first identity public key, Kn_D and a first identity certificate to the second security context, wherein the first identity public key, K_1C pu_b and the first identity private key, Kn_D pnv are a cryptographic key pair and the first identity certificate comprises information identifying K_1C put, and is cryptographically signed by a manufacturer private key K^p™; and

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 receive a second identity public key, K₂id _pub, and a second identity certificate from the second security context, the second identity certificate comprising information identifying K₂io _pub and being cryptographically signed by the manufacturer private key Kmanp™;

the first processor is further configured to verify the second identity certificate using the manufacturer public key Kmanpub.

In an embodiment, the first transceiver is further configured to;

receive information relating to the current configuration of the second security context, wherein the information is cryptographically signed with a second identity private key, K₂id pm, wherein the second identity public key, K_2C pu_b and the second identity private key, KaDpriv are a cryptographic key pair; and wherein the first processor is further configured to:

verify the signature using the second identity public key, K_2Ü pu_b; validate that the configuration of the second security context meets the security requirements of the tenant and that the second security context is configured to enforce the policies contained in the ACL.

In an embodiment, the first processor is further configured to:

generate a second cryptographic key pair and a second cryptographic certificate in the first security context, the second cryptographic key pair comprising a second public key, K_tenant__sjgnpub, and a second private key, K_tenant__sigBprtv and the second cryptographic certificate comprising information from which the origin of the second public key K_tenant__S|_gnpub can be identified;

cryptographically sign the second cryptographic certificate with the first identity private key, K_1Cprtv;

wherein the first transceiver is further configured to send the second public key Ktenant-sign_pub and the signed second cryptographic certificate to the second security context

In a further aspect of the present invention, there is provided a cryptographic device, comprising a first security context, comprising:

a first processor, configured to generate a use credential comprising:

information from which data corresponding to the use credential can be identified;

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 information from which an expiry of the use credential can be determined;

a first transceiver, configured to send the use credential and information from which the origin of the use credential can be validated to a second security context.

In an embodiment, the use credential is a use credential.

In an embodiment, the data comprises a cryptographic key, KenantIn an embodiment, the first security context further comprises:

a first device memory, storing a private key K_tenant__signpriv;

wherein the first processor is configured to cryptographically sign the use credential with the private key K_tenant__signprlv.

In an embodiment, the use credential is a use certificate.

In an embodiment, the information from which the expiry of the use credential can be determined comprises:

an expiry time;

information identifying a reference time source.

In an embodiment, the information from which the cryptographic key Kenam corresponding to the use credential can be identified is a hash of K^antIn a further aspect of the present invention, there is provided a cryptographic device comprising a second security context, for cooperation with a device or devices comprising a first security context, the cryptographic device comprising:

a processor, configured to perform cryptographic operations, the processor being configured to:

generate a first cryptographic key pair and a first cryptographic certificate, the first cryptographic key pair comprising a first public key, Kblob pub, and a first private key, Kblob pnv and the first cryptographic certificate comprising information from which the origin of the first public key Kblob pub can be validated;

a transceiver, configured to:

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 send the first public key K_Bioe pub and the first cryptographic certificate to a first security context; and receive encrypted data and a corresponding access control list, and information from which the origin of the data can be validated from the first security context;

the processor further configured to:

validate the origin of the data;

decrypt the encrypted data and corresponding access control list using the first private key Kblobpôv·

In an embodiment, the use credential is a use certificate.

In an embodiment, the data comprises a cryptographic key, K^ant·

In an embodiment, the processor is further configured to:

re-encrypt the cryptographic key Ktenam with a further cryptographic key, wherein the further cryptographic key cannot leave the second security context.

In an embodiment, the device further comprises a device memory, storing a second identity private key, K_2C pov and wherein the transceiver is further configured to:

send a second identity public key, K₂id pub. and a second identity certificate to the first security context, wherein the second identity public key, K₂|_D pu_band the second identity private key, Κ_2β pdv are a cryptographic key pair and the second identity certificate comprises information identifying K_2)D pub and is cryptographically signed by a manufacturer private key Kman p^; and receive a first identity public key, K_w pu_b, and a first identity certificate from the first security context, the first identity certificate comprising information identifying Ki_ID pub and being cryptographically signed by the manufacturer private key Kmen priv the processor is further configured to;

verify the first identity certificate using the manufacturer public key K^n pub

In an embodiment, the processor is further configured to:

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 generate information relating to the current configuration of the second security context;

cryptographically sign the information with the second identity private key, K_ao pnv, and wherein the transceiver is further configured to:

send the information and signature to the first security context.

In an embodiment, the processor is further configured to:

generate the first cryptographic certificate comprising the information relating to the current configuration of the second security context and sign the first cryptographic certificate with the second identity private key, K₂|_Dp<iv·

In an embodiment, the transceiver is further configured to:

receive a second public key K_tenant__Sj_gnpub and a signed second cryptographic certificate, the second cryptographic certificate comprising information from which the origin of the second public key Ktenant-sign_pub⁰³¹¹ be validated, from the first security context;

and wherein the processor is further configured to:

validate the origin ofthe second public key K_tenant-signp_Ut)

In a further aspect of the present invention, there is provided a device, comprising:

a device memory, storing:

encrypted data;

an access control list corresponding to the data, the access control list specifying that a valid use credential must be presented in order to grant a first type of use of the data, and specifying that the use credential must comprise information from which the expiry of the use credential can be determined and must be unexpired in order to grant a first type of use of the data; and information from which the origin of the data can be identified.

In an embodiment, the use credential is a use certificate.

In an embodiment, the device is a cryptographic device comprising a second security context, for cooperation with a device or devices comprising a first security context.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

In an embodiment, the data comprises a cryptographic key, Κ^,,^.

The access control list may specify that the use credential is a use certificate which must be signed by a private key K_tenant-sign_pr|_V in order to grant use of the cryptographic key, K^ant.

The access control list may specify that the use credential must comprise information from which the cryptographic key Ktenant corresponding to the use credential can be identified.

In an embodiment, the encrypted cryptographic key Ktenant is encrypted by a key which cannot leave the second security context.

In an embodiment, the device further comprises:

a processor, configured to validate a received use credential with respect to the access control list, and validate that the use credential has not expired;

grant the first type of use of the cryptographic key, Ktenant, in the second security context, on the condition that the use credential is valid and not expired.

In an embodiment, there is provided a method of data transfer from a tenant to a service provider comprising encrypting the data with a public key of a key pair generated by a secure device within the service provider system. The data thus cannot be accessed by the service provider during transmission.

The data is generated with a corresponding access control list, which specifies that a valid certificate must be presented in order to grant a particular use of the data once stored. The tenant can thus retain control of the use of the data even though it has been transferred out of the tenant system.

In an embodiment, there is provided a method of controlling use of data securely stored in the service provider system comprising issuing a use credential having an expiry time to the party requesting use of the data. The use credential must be validated before use of the stored data is granted. This enables the tenant to grant use of the stored data for a limited time period.

Date Reçue/Date Received 2021-06-25

15a

According to an aspect of the present invention there is provided a cryptographic device, comprising a first security context, comprising:

a first processor, configured to generate a use credential comprising:

information from which data corresponding to the use credential can be identified;

information from which an expiry ofthe use credential can be determined; a first transceiver, configured to send the use credential and information from which the origin ofthe use credential can be validated.

In some embodiments, the data comprises a cryptographic key, Ktenant.

In some embodiments, the first security context further comprises:

a first device memory, storing a private key K_tenant-sign_prlv;

wherein the use credential is a use certificate and the first processor is configured to cryptographically sign the use certificate with a private key K_tenant__signprjv.

In some embodiments, the information from which the expiry ofthe use credential can be determined comprises:

an expiry time;

information identifying a reference time source.

In some embodiments, the information from which the cryptographic key Ktenant corresponding to the use credential can be identified is a hash of Ktenant.

According to another aspect ofthe present invention there is provided a method of controlling use of data, the data being stored in a service provider system in a manner which is accessible to a trusted second security context in the service provider system but secure from the rest ofthe service provider system, wherein an access control list specifying that a valid use credential must be presented in order to grant a first type of use of the data is stored with the data, the method comprising:

Date Reçue/Date Received 2021-06-25

15b generating a use credential in a first security context, wherein the use credential comprises:

information from which the data corresponding to the use credential can be identified;

information from which the expiry of the use credential can be determined;

issuing the use credential and information from which the origin of the use credential can be validated;

validating the use credential with respect to the access control list, and validating 10 that the use credential has not expired at the second security context;

granting the first type of use of the data, in the second security context, on the condition that the use credential is valid and not expired.

According to another aspect of the present invention there is provided a carrier medium 15 comprising computer readable code configured to cause a computer to perform the method as described herein.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

In this specification, the term security context refers to one or more security appliances (for example HSMs), or partitions of a security appliance, which share at least one private key and are configured to safeguard and perform a set of cryptographic functions.

In this specification, the term cryptographic key refers to a block of raw cryptographic material for use in cryptographic operations. An access control list corresponding to the key comprises information relating to a set of permissions describing the operations the key material may be used for, for example encryption, decryption or storage, and any credentials that must be provided to enable permissions. Also associated with the key may be information relating to the key type which comprises data identifying the type of key, including information identifying, for example, the algorithms the key can be used with and its length, e.g. the Advanced Encryption Standard (AES) algorithm with a key of length 256 bits, or the RSA algorithm with a key length of 2048 bits.

In this specification, the term “verify may be used to refer to a method of checking a cryptographic signature. The term “validate” may be used to refer to a method of checking that data is as expected, or to a method of checking both that a signature is cored and that the data is as expected.

In this specification, the term “access control list” refers to one or more permissions attached to an object. The permissions specify which operations are allowed on the object, and the conditions and/or credentials required for the operation to be granted.

In the methods described in this specification, public keys may be stored within the first security context or the second security context, or on untrusted mediums outside of the first security context or the second security context if the public key is integrity proteded, for example signed by the identity key of the first security context or the second security context. This means that if the public key is tampered with, it will be detected.

In the methods described in this specification, information sent between the first security context and the second security context may be validated by use of a cryptographic certificate. For example, the information may be signed by a private

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 signing key belonging to the sender, and the signature sent together with the information to the receiver.

The methods described in this specification may be computer implemented methods.

Since some methods in accordance with embodiments can be implemented by software, some embodiments encompass computer code provided to a general purpose computer on any suitable carrier medium. The carrier medium can comprise any non-transient storage medium such as a floppy disk, a CD ROM, a magnetic device or a programmable memory device, or any transient medium such as any signal for example an electrical, optical or microwave signal.

Brief description of the Figures

Devices and methods in accordance with non-limiting embodiments will now be described with reference to the accompanying figures in which:

Figure 1(a) is a schematic illustration of a network comprising a tenant system comprising a cryptographic device in accordance with an embodiment of the present invention and a service provider system comprising a cryptographic device in accordance with an embodiment of the present invention;

Figure 1(b) shows a schematic illustration of a first security context in accordance with an embodiment of the present invention and a second security context in accordance with an embodiment of the present invention;

Figure 2(a) is a flow chart showing a method of establishing trust between the first security context and the second security context, which is part of a method of cryptographic key transfer according to an embodiment of the present invention;

Figure 2(b) is a flow chart showing a method of validating the configuration of the second security context at the first security context, which is part of a method of cryptographic key transfer according to an embodiment of the present invention;

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

Figure 3 is a schematic illustration of a first security context in accordance with an embodiment of the present invention and a second security context in accordance with an embodiment of the present invention, after trust has been established, and after the relevant keys have been generated in each security context;

Figure 4(a) is a flow chart showing a method of transferring a signing key, Kt_enant._Sig_nfrom the first security context to the second security context, which is part of a method of cryptographic key transfer according to an embodiment of the present invention;

Figure 4(b) is an illustration of a method of tenant registration, which is part of a method of cryptographic key transfer according to an embodiment of the present invention;

Figure 5 is a schematic illustration of a first security context in accordance with an embodiment of the present invention and a second security context in accordance with an embodiment of the present invention after the second public key, ^_enam._Sian pub, has been exchanged during the data transfer process;

Figure 6(a) is a flow chart showing a method of transferring data from the first security context to the second security context in accordance with an embodiment of the present invention;

Figure 6(b) is a flow chart showing further steps of a method of transferring data from the first security context to the second security context in accordance with an embodiment of the present invention;

Figure 7(a) is an illustration of a method of key enrolment, which is part of a method of cryptographic key transfer according to an embodiment of the present invention;

Figure 7(b) is a flow chart of a method of cryptographic key transfer according to an embodiment of the present invention;

Figure 8(a) is a schematic illustration of a first security context in accordance with an embodiment of the present invention and a second security context in accordance with

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 an embodiment of the present invention after the tenant key Ktenant has been imported to the second security context;

Figure 8(b) is a schematic illustration of a time source which may be hosted by the service provider, the tenant or a separate third-party;

Figure 9 is a flow chart showing a method of registering a time source with the second security context;

Figure 10 is a flow chart showing a method of registering a time source with the first security context;

Figure 11 is a flow chart showing a method of controlling use of data according to an embodiment of the present invention;

Figure 12 is a flow chart showing a method of generating a use certificate at the first security context, which is part of a method of controlling use of a cryptographic key, tenant according to an embodiment of the present invention;

Figure 13 is an illustration of a method of controlling use of a cryptographic key, Kt_enant according to an embodiment of the present invention;

Figure 14 is a flow chart of a method of controlling use of a cryptographic key, K_tenant according to an embodiment of the present invention.

Detailed description

Figure 1(a) is a schematic illustration of a tenant system 1 comprising a cryptographic device in accordance with an embodiment of the present invention and a service provider system 3 comprising a cryptographic device in accordance with another embodiment of the present invention.

The service provider may be a cloud service provider for example. The service provider provides storage of data such as cryptographic keys and secure cryptographic processing of data to one or more tenants. For example, the tenants may use the

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 cryptographic infrastructure of the service provider system 3 for applications such as payments, security and regulation. The service provider system 3 comprises a service provider application server, configured to perform one or more applications such as these.

In order to use these services, the tenant system 1 provides a cryptographic key Kt_enant to the service provider system 3. The cryptographic key K_tenant is securely stored at the service provider system 3 for use in such applications.

The tenant system 1 comprises a first security context 5 and the service provider system 3 comprises a second security context 7. A security context may be a single security device, for example a hardware security module, HSM. Alternatively, it may be two or more security devices, or a partition of a security device. The term security context is used here to refer to the device, devices, or partition of a device which form a single security context, i.e. which share at least one private key and are configured to safeguard and perform a set of cryptographic functions. The first security context 5 is secured from the rest of the tenant system. The second security context 7 is secured from the rest of the service provider system.

The tenant cryptographic key Kenant is provided to the second security context 7 in the service provider system 3. The tenant cryptographic key Ktenant is then either stored within the second security context 7 or is encrypted with a key that cannot leave the second security context before being stored elsewhere in the service provider system 3.

Before providing the cryptographic key Ktenant to the second security context 7, the tenant can authenticate and validate the second security context 7 from a generation certificate provided by the second security context 7. The generation certificate may have been generated at the time of manufacture of the device or devices which are part of the second security context 7 for example. The tenant trusts the manufacturer but not the service provider. The generation certificate authenticates that the second security context 7 was manufactured by the trusted manufacturer, and thus can be trusted. Furthermore, the parameters and state of the second security context 7 can be validated, for example from information contained in the generation certificate, and in a further configuration certificate.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

A generation certificate comprises static information that is valid for the entire lifetime of the appliance. A configuration certificate comprises information about the current configuration of the device. Configuration information is valid only at the time of generation of the configuration certificate, and may change at a later stage.

The first security context 5 thus establishes trust with the second security context, before key transfer operations begin.

The tenant cryptographic key Ktenant is encrypted with the public half of an asymmetric key pair generated at the second security context 7, before being transferred to the second security context 7. A first cryptographic certificate is issued with the asymmetric key pair generated at the second security context 7. In an embodiment, the first cryptographic certificate validates that the asymmetric key pair was generated in the second security context 7, that the private half ofthe asymmetric key pair is ephemeral and that the private half of the asymmetric key pair cannot leave the second security context 7. This enables the tenant cryptographic key Ktenant to be transferred to the service provider system 3 in a manner which is secure from attackers and from the service provider itself, i.e. from the rest of the service provider system 3 which is outside of the second security context 7, for example the application server.

In an embodiment, the tenant cryptographic key K_tenant is then stored in the second security context 7. Alternatively, the tenant cryptographic key K_te_nant is stored elsewhere in the service provider system, for example in the application server, encrypted by a key which cannot leave the second security context 7.

An access control list (ACL) corresponding to the tenant cryptographic key Ktenant is also generated with the tenant cryptographic key Ktenam in the first security context 5. The ACL is also transferred to the second security context 7 with the tenant cryptographic key K_tenanl. The ACL is stored with the tenant cryptographic key K_tenant. The first security context 5 has established trust with the second security context 7, and thus knows that the second security context 7 will enforce the policies contained in the ACL. The ACL thus allows the tenant to retain control over the key, even once it has been transferred to the second security context 7.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

In an embodiment, the ACL specifies that the tenant cryptographic key Ktenant can only be stored inside the second security context 7. In an alternative embodiment, the ACL contains a restriction that the tenant cryptographic key Ktenant can only be stored on the condition that it is encrypted for storage by a key which cannot leave the second security context 7. This ensures that the tenant cryptographic key Ktenant is inaccessible to third parties and to the service provider.

The ACL may specify that the tenant cryptographic key Ktenant can only be stored in non-volatile memory being resistant to tamper by a third party.

Furthermore, the ACL specifies that a valid use credential, for example a use certificate, must be presented in order to grant one or more types of use of the tenant cryptographic key Ktenant· For example, the ACL may require presentation of a certificate signed by the private half of an asymmetric key owned by the tenant to grant performance of certain cryptographic operations using the tenant key K_tenant· This ensures that the particular type of operations using the key cannot be used if not authorised by the tenant

Although the tenant cryptographic key Kt_enant is stored in the service provider system, the service provider, i.e. the rest of the service provider system 3 which is outside of the second security context 7, cannot access the key, and the key cannot be used without authorisation from the tenant. This protects the tenant cryptographic key Kenam from a malicious service provider. It also protects the tenant cryptographic key Κ_θη3η1from security agencies which have jurisdiction over the service provider but not the tenant, for example.

It also allows multiple tenants to use the same cryptographic infrastructure in the service provider. Multiple tenants can store keys in the same security device, as each tenant cryptographic key Kt_enam is inaccessible to the other tenants, and cannot be used without authorisation from the corresponding tenant.

The access control list may specify that the use credential must comprise information from which the expiry of the use credential can be determined and must be unexpired in order to grant use of the cryptographic key, K_tenant. The tenant is thus able to specify,

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 in the use credential, an expiry period for its key, after which the key may not be used until further authorization is provided.

The expiry time can be calculated with reference to a reference time source 2 that is trusted by both the first security context 5 and the second security context 7. The reference time source 2 may be hosted by the service provider, the tenant or a separate third-party. Figure 1(a) shows a schematic illustration of an embodiment in which the reference time source 2 is hosted by a third party.

Although the above description relates to transfer and storage of a tenant cryptographic key, any form of data may be transferred and stored in the same manner. The method of data transfer from a tenant to a service provider comprises encrypting the data with a public key of a key pair generated by a secure device within the service provider system. The data thus cannot be accessed by the service provider during transmission. The encrypted data is cryptographically signed before transfer, to ensure authenticity and integrity of the data.

The data is generated with a corresponding access control list, which specifies that a valid certificate must be presented in order to grant a particular use of the data once stored. The tenant can thus retain control of the use of the data even though it has been transferred out of the tenant system.

A method of controlling use of data securely stored in the service provider system comprises issuing a use credential having an expiry time to the party requesting use of the data. The use credential must be validated before use of the stored data is granted. This enables the tenant to grant use of the stored data for a limited time period.

Figure 1(b) shows a schematic illustration of a first security context 5 in accordance with an embodiment of the present invention and a second security context 7 in accordance with another embodiment of the present invention.

The first security context 5 may be a single security appliance, for example a hardware security module, HSM. Alternatively, the first security context 5 may be two or more security appliances, or a partition of a security appliance. The first security context 5 could be a low-power, low performance, low cost HSM for example.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

The second security context 7 may be a single security appliance, for example a hardware security module. Alternatively, the second security context 7 may be two or more security appliances, or a partition of a security appliance. The second security context 7 may comprise a cluster of high-performance HSMs.

The first security context 5 and the second security context 7 may thus each comprise one or more tamper resistant cryptographic appliances or a partition of a tamper resistant cryptographic appliance.

The first security context 5 comprises a first device memory 9. The first device memory 9 is configured to store cryptographic information such as keys, key pairs and certificates. The first device memory 9 may include any form of non-volatile device memory such as flash, optical disks or magnetic hard drives for example. The first security context 5 also comprises volatile memory.

The first device memory 9 may be physically secure and may be resistant to tamper by a third party, for example by the inclusion of physical security such as a membrane that covers the entire device, that cannot be removed without destroying the underlying physical hardware, thus making it un-usable.

A unique asymmetric identity key K_1ID is stored in the first device memory 9, with a corresponding signed generation certificate {CicÏKman ρην is a signing key used to prove the origin of data and authenticity. The generation certificate Cud may describe the public parameters of the key, for example, the generation certificate Cw may include information relating to the type of the key, and its length. The generation certificate Cud comprises information which authenticates that the identity key Kw was generated in the first security context 5. For example, the generation certificate Cn_Dmay comprise the hash of the public half of Kn_D and be signed by the private half of a manufacturer asymmetric key, KmanpuvThe generation certificate Cn_D may also include state information, for example, information relating to a unique identification of the device, information identifying the manufacturer, the hardware version used, the software type used, the serial number of the unit and the supported features/functionality of the model. The generation

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 certificate Cæ is signed by a manufacturer trusted by both the first security context 5 and the second security context 7. The generation certificate is cryptographically signed with the private half of a manufacturer asymmetric key, K_TOn pr>v The manufacturer may be a third party who manufactured the security appliance(s) which forms the first security context and the security appliance(s) which forms the second security context. The public half of the trusted manufacturer key Kman pub is also stored in the first device memory 9, or may be stored outside of the first security context 5 in an integrity protected manner.

The second security context 7 comprises a second device memory 11. The second device memory 11 is configured to store cryptographic information such as keys, key pairs and certificates. The second device memory 11 may include any form of nonvolatile device memory such as flash, optical disks or magnetic hard drives for example. The second security context 5 also comprises volatile memory.

The device memory may be physically secure and may be resistant to tamper by a third party, for example by the inclusion of physical security such as a membrane that covers the entire device, that cannot be removed without destroying the underlying physical hardware, thus making it un-usable.

A unique asymmetric identity key Kao is stored in the second device memory 11, with a corresponding signed generation certificate {C_2E}_Kman pnv- Kao is a signing key used to prove the origin of data and authenticity. The generation certificate Cao may describe the public parameters of the key, for example, the generation certificate C_2E may include information relating to the type of the key, and its length. The generation certificate CaD comprises information which authenticates that the identity key Κ_2Ε was generated in the second security context 7. For example, the generation certificate C₂d may include a hash of the public half of K_2D, and be signed by the private half of a manufacturer asymmetric key, K^p™.

The generation certificate Cao may also include state information, for example, information relating to a unique identification of the device, information identifying the manufacturer, the hardware version, the software type used, the serial number of the unit and the supported features/functionality of the model. The generation certificate Cao is signed by the trusted manufacturer. The generation certificate is

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 cryptographically signed with the private half of a manufacturer asymmetric key, Kma_npnv* The public half of the trusted manufacturer key K_man is also stored in the second device memory 11, or may be stored outside of the second security context 7 in an integrity protected manner.

In both the first security context 5 and the second security context 7, cryptographic keys are stored in the device memory in a secure, tamper-proof format.

The first security context 5 and second security context 7 can be verifiably identified using cryptographic certificates, the signed generation certificates {Cn_D} Kman and {Cao} Kman priv. which may be generated at the time of manufacture. Each is thus able to securely store their identity in a way that means they cannot be imitated. Identifying the first security context 5 and second security context 7 allows for the origin of the device or devices in each security context to be ascertained. Each device contains the unique asymmetric identity key K|_D generated in the factory when it was manufactured, for example. Each component also contains the key-generation certificate for the K_ID which is signed using an asymmetric key known only to the manufacturer. The public half of the manufacturer key can be used as the root of trust to authenticate genuine appliances.

Furthermore, the parameters and state of the first security context and second security context can be validated in a non-repudiable manner, from information contained in the generation certificate, and through exchange of further, configuration, certificates.

In an embodiment, the first security context 5 and second security context 7 are each configured to generate a configuration certificate, Civ and Czv respectively, containing information about the current configuration of the devices set by the tenant and service provider. The configuration information cannot be included in the generation certificates, as these are generated at the time of manufacture, before the devices are distributed to the tenant and service provider and configured.

Thus once the generation certificates have been exchanged, and trust formed between the two security contexts, further information that relates to the dynamic configuration can be exchanged through further signed certificate data transfer. The configuration certificates, C_1V and C_2V respectively, are signed by the private half of the unique

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 asymmetric identity key of the corresponding security context. The configuration certificate is signed by Kopdv to verify authenticity, as Kcpriv is now trusted by the other security context. The data contained in the configuration certificate may pertain to the administrator deployment options such as security settings, software version, which trusted time source is used or whether the HSM believes there has been an attempt to tamper with it for example.

The origin and state of a service define enough information by which other services can place their trust. This information is exchanged in the generation and configuration certificates.

The second security context is trusted to enforce the rules specified in an ACL provided to the second security context and trusted to update its state certificates accurately. In an embodiment, if the second security context cannot enforce a rule to the specified level contained within the ACL then it will not perform an operation and will not advertise itself or the keys that it has created as supporting said rules.

The identity keys of the first and second security context may be installed at manufacture, and the hash of these keys signed by a manufacturer key K_nan priv, and stored with the identity key, thus proving its provenance.

By being able to cryptographically identify services from trusted sources, a tenant is able to exchange its cryptographic key with a hosted service over a communication channel with a high assurance that all data is protected and not recoverable by any third-party, including the hosting service provider. By ascertaining that the service provider’s appliance(s) in the second security context 7 is trusted, from the generation certificate, and that its configuration meets the tenant’s security policy, from the configuration certificate, the tenant can transfer their key knowing that the key will be stored in the manner specified. The ACL sent with the key includes policies specifying how the key is to be stored and used. The second security context 7 will then enforce the policies. The tenant can trust that the second security context 7 will enforce the policies as it has established trust with the second security context 7. Establishing trust allows secure transfer and the ACL allows the second security context 7 to apply a policy once it is in possession of the key.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

The first security context 5 further comprises a transceiver 13. The transceiver 13 is configured to transmit and receive data packets. The data packets may be transmitted from and received at the first transceiver 13, for example via an Internet connection or a directly wired connection between the first security context 5 and the second security context 7. This communication link may be untrusted, however the key transfer protocol described in relation to Figure 6(a) below provides protection of the key from attackers.

The first security context 5 further comprises a first processor 17. The first processor 17 is configured to perform cryptographic operations, such as generation of cryptographic keys and asymmetric cryptographic key pairs, generation of certificates corresponding to a cryptographic key or asymmetric cryptographic key pair, generation of access control lists corresponding to a cryptographic key, generation of use certificates corresponding to a cryptographic key, encryption of an object with a cryptographic key which is stored in the first device memory 9, decryption of an encrypted object with a cryptographic key which is stored in the first device memory 9, cryptographically signing an object with a cryptographic key which is stored in the first device memory 9, verification of a cryptographic signature and validation of an object based on information stored in the first device memory 9. The first processor 17 may be physically secure.

In an embodiment, the first security context 5 comprises a main processor to perform non-cryptographic operations and the first processor 17 is a co-processor, i.e. a separate component from the main processor configured to perform only the cryptographic operations. Alternatively, the first processor 17 may be the main processor.

The generation of cryptographic keys and asymmetric cryptographic key pairs may comprise generation of random numbers. The first security context 5 may further comprise a random entropy source, for use in random number generation.

The second security context 7 further comprises a transceiver 15. The second transceiver 15 is configured to transmit and receive data packets. The data packets may be transmitted from and received at the second transceiver 15, for example via a wireless Internet connection or a directly wired connection between the first security context 5 and the second security context 7.

Date Reçue/Date Received 2021-06-25

WO 2017/134445

PCT/GB2017/050264

The second security context 7 further comprises a second processor 19. The second processor 19 is configured to perform cryptographic operations, such as generation of cryptographic keys and asymmetric cryptographic key pairs, generation of certificates corresponding to a cryptographic key or asymmetric cryptographic key pair, generation of access control lists corresponding to a cryptographic key, generation of use certificates corresponding to a cryptographic key, encryption of an object with a cryptographic key which is stored in the second device memory 11, decryption of an encrypted object with a cryptographic key which is stored in the second device memory 11, cryptographically signing an object with a cryptographic key which is stored in the second device memory 11, verification of a cryptographic signature and validation of an object based on information stored in the second device memory 11. The second processor 19 may be physically secure.

In an embodiment, the first security context 5 comprises a main processor to perform non-cryptographic operations and the first processor 17 is a co-processor, i.e. a separate component from the main processor configured to perform only the cryptographic operations. Alternatively, the first processor 17 may be the main processor.

The generation of cryptographic keys and asymmetric cryptographic key pairs may comprise generation of random numbers. The second security context 7 may further comprise a random entropy source, for use in random number generation.

A HSM device such as may be used as part of the first security context 5 or second security context 7 may comprise a device memory, processor, transceiver and random entropy source as described above. The HSM may comprise both physical and nonphysical security properties. Non-physical security properties include the use of encryption, i.e. the inclusion in the device of software or a physical component configured to perform encryption of stored data. Physical properties can include tamper switches triggered by physical access, and a tamper proof membrane surrounding the physical boundary of the device.

The cryptographic asymmetric key pairs discussed in this application may be any asymmetric key pair type that supports signing and verification. For example, each of

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 the manufacturer key pair Kman, the fist identity key identity key Ku_D, second identity key pair K2©, time source identity key pair Kts© and signing key pair IÇ_enant sign may be any one of an RSA, a DSA, or a ECDSA key pair for example, where RSA or DSA are the algorithm for signing and verification.

For example, to generate an RSA key pair, a generation operation may be performed by the first processor 17, second processor 19 or third processor 41 to generate an output key pair which can be used by the RSA algorithm to sign data.

A random entropy source may be used to generate random numbers, which in turn are used by the first processor 17, second processor 19 or third processor 41 to generate cryptographic keys and cryptographic key pairs.

The cryptographic asymmetric key pairs in this application may be any asymmetric cryptographic key pair type that supports encryption and decryption. For example, the first cryptographic key pair Kblob may be an RSA key pair or a key pair that could be used in an Integrated Encryption Scheme (IES) algorithm.

The transfer of cryptographic certificates and cryptographic keys between the first security context 5 and the second security context 7 described in the following may occur over a secure, authenticated channel between the first security context 5 and the second security context 7, which is provided and controlled by the service provider. The secure channel provided by the service provider may be provided using a load balancer or firewall appliance. Use of such infrastructure mitigates attacks such as denial of service.

Although the channel is secure from third parties, it is open to attack by the service provider, and therefore is not trusted by the tenant for transfer of high value cryptographic keys. Thus security is enforced by the tenant by the encryption of Κ(_βη8ηιand subsequent signing of the encrypted data sent over the channel.

Figure 2(a) is a flow chart showing a method of establishing trust between the first security context 5 and the second security context 7. The method of establishing trust is part of a method of cryptographic key transfer according to an embodiment of the present invention. The method of establishing trust may be performed before the

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 cryptographic key K^nant is generated for example, or may be performed after the cryptographic key is generated but before any exchange of cryptographic keys between the first security context 5 and the second security context 7 for example.

In step S201, the public half of the identity key of the first security context, Ku_D pub. and the generation certificate {Cæ} Kman pnv, are sent from the first security context 5 to the second security context 7. The first transceiver 13 in the first security context 5 is configured to send the public half of the identity key of the first security context, Kud p^, and the generation certificate {Cæ} Kman p™, to the second transceiver 15 in the second security context 7. Information relating to the state of the device(s) in the first security context 5 may be sent in the same message. The information relating to the state of the device(s) is also included in the generation certificate in this case, and can be used by the receiver to validate the state information contained in the message.

The generation certificate is immutable, thus only contains information that is available at manufacture time. Information relating to the current configuration of the device, for example, the IP address or tamper state, may be included in the configuration certificate. The configuration information may be sent at the same time as the generation certificate, or after the generation certificate. However, the generation certificate is generated and signed at the time of manufacture, whereas the configuration certificate is generated at the time of key transfer, and is signed by K_1C_ priu. The configuration certificate can be verified only after the generation certificate has been verified.

In step S202, the signature of the generation certificate {Cud} Kman pnv is verified at the second security context 7. The second processor 19 in the second security context 7 is configured to verify the signature of the generation certificate {Cud } Kman pnv- The generation certificate is verified using the public half of the trusted manufacturer key Kman _pub which is stored in the device memory 11. The second processor 19 is configured to perform a signature verification algorithm that, given the signed message {Ci id } Kman pnv and the public key _pUb either accepts or rejects the message's claim to authenticity.

The authenticity of the public half of the identity key of the first security context, K_{1C pub}is validated at the second security context 7. The second processor 19 in the second

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 security context 7 is configured to validate the public half of the identity key of the first security context, Ku_{D pU}b- In an embodiment in which the generation certificate Cu_Dcomprises a hash of the public half of the identity key of the first security context, Kæ pub, the authenticity is validated by calculating the hash the public half of the identity key of the first security context, K_1C pub, and validating that it matches that contained in its generation certificate CudIn an embodiment, the generation certificate Cud comprises the hash of the public half of Κι ο and is signed by the private half of the manufacturer asymmetric key, Kman p™ Step S202 comprises: calculating the hash of the received data sent in the message, in this case the hash of Ki_C pubi inputting the received signature into the verification algorithm, which may be a cryptographic operation, together with the public half of the manufacturer key and verifying the output; comparing the calculated hash to that contained in the output of the verification algorithm to determine whether they are the same.

In step S203, if state information is included in the message, the second security context 7 validates that the state of the device(s) meets requirements. Alternatively, state information is not included in the generation certificate, and the second security context 7 does not validate the state of the first security context 5. The second security context 7 does not transfer any secure information to the first security context 5, thus it is not necessary for the state information of the first security context 5 to be validated.

If the signature is verified and state information is validated, the public half of the identity key of the first security context, Kæ _pub, is stored in the second device memory 11 of the second security context 7. Alternatively, the public half of the identity key of the first security context, Ku_D pub can be integrity protected by the second security context 7 and stored in untrusted storage outside of the second security context 7. The second security context 7 may sign the public half of the identity key of the first security context, KnDpub. and data indicating that this is a public key from a trusted source, using K2© pnv for storage. Alternatively, it may encrypt them using another secret key. In these cases, an untrusted device memory can be used, which often has a larger capacity than the device memory within the second security context 7, whilst trust in the key is maintained.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

If the signature is not verified or the state information is not validated, an error is returned to the first security context, for example a message is sent stating “AccessDenied”. At this point, the communication between the first security context 5 and the second security context 7 is terminated.

In step S204, the public half of the identity key of the second security context, K^d pub, and the generation certificate {C_2)D} Kman _Priv, are sent from the second security context 7 to the first security context 5. The second transceiver 15 in the second security context 7 is configured to send the public half of the identity key of the second security context, K₂id pub. and the generation certificate {C₂id } Kman pdv. to the first transceiver 13 in the first security context 5. Information relating to the state of the device(s) in the second security context 7 may be sent in the same message. The information relating to the state of the device(s) will also be included in the generation certificate in this case, in order to validate the state information. Again, the generation certificate is immutable, thus only contains information that is available at manufacture time. Information relating to the current configuration of the device, for example, the IP address or tamper state, may be included in the configuration certificate. The configuration information may be sent at the same time as the generation certificate, or after the generation certificate, for example in step S604 as part of the first cryptographic certificate Cblob However, the generation certificate is generated and signed at the time of manufacture, whereas the configuration certificate is generated at the time of key transfer, and is signed by KziD-priv- The configuration certificate can be verified only after the generation certificate has been verified.

In step S205, the signature of the generation certificate {C₂io } Kman p<iv is verified at the first security context 5. The first processor 17 in the first security context 5 is configured to verify the generation certificate {C21D} Kman pnv- The signature is verified using the public half of the trusted manufacturer key Kman pub stored in the device memory 9. The first processor 17 is configured to perform a signature verification algorithm that, given the signed message {C_2tD} Kman priv and the public key Km_an either accepts or rejects the message's claim to authenticity. This allows the first security context 5 to validate the manufacturer device in the second security context 7, by verifying the signature.

The authenticity of the public half of the identity key of the second security context, K_2Cpub is validated at the first security context 5. The first processor 17 in the first security

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 context 5 is configured to validate the public half of the identity key of the second security context, K₂id _pub. In an embodiment in which the generation certificate Cao comprises a hash of the public half of the identity key of the second security context, K21D pub, the authenticity is validated by calculating the hash the public half of the identity key of the second security context, Kao pub, and validating that it matches that contained in its generation certificate Cao

In step S206, if state information is included in the message, the first security context 5 validates that the state of the device(s) meets requirements.

If the signature is verified and state information is validated, the public half of the identity key of the second security context, Ka_D is stored in the first device memory 9 of the first security context 5. Alternatively, the public half of the identity key of the second security context, Ka_{D pu}b can be integrity protected by the first security context 5 and stored in untrusted storage outside of the first security context 5. The first security context 5 may sign the public half of the identity key of the second security context, K_2IDpub, and data indicating that this is a public key from a trusted source, using K_{1C priv} for storage. Alternatively, it may encrypt them using another secret key. In these cases, an untrusted device memory can be used, which often has a larger capacity than the device memory within the first security context 5, whilst trust in the key is maintained.

If the signature is not verified or the state information is not validated, an error is returned to the second security context 7, for example a message is sent stating “AccessDenied”. At this point, the communication between the first security context 5 and the second security context 7 is terminated.

Figure 2(b) is a flow chart showing a method of validating the configuration of the second security context 7 at the first security context 5, which is part of a method of cryptographic key transfer according to an embodiment of the present invention. In this method, the configuration of the second security context 7 can be validated. In an embodiment, a similar method can also be used to validate the configuration of the first security context 5 at the second security context 7. Alternatively, the second security context 7 does not validate the configuration of the first security context 5. The second security context 7 does not transfer any secure information to the first security context

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

5, thus it is not necessary for the configuration of the first security context 5 to be validated.

Configuration information is generated by the second security context 7 in step S211. The configuration information may include information relating to the specific configuration applied by the administrator. This may include information relating to the cryptographic operations that are supported, the encryption keys that are used, and/or the version of software of the unit, for example. The configuration information may comprise information indicating that the first security context 5 has been configured to use the AES algorithm to encrypt tenant keys when not inside the second security context 7 for example. The configuration information may further comprise information relating to the administrator deployment options such as security settings, which trusted time source is used or whether the second security context 7 believes that there has been an attempt to tamper with the device.

In step S212, the configuration information is signed by the private half of the identity key of the second security context, K₂|_D producing a configuration certificate C_2V. The private half of the identity key of the second security context, K₂|_{D pnv} is stored in the second security context 7 and cannot be accessed by an administrator. Signing the configuration certificate C₂v with the private half of the identity key of the second security context, K_{2D priv} means that the configuration information cannot be subverted.

In step S213, the configuration information and configuration certificate Czv is sent from the second security context 7 to the first security context 5. This can be sent at the same time as K_2D pub and {C₂id } _Kman priv, in step S204 for example. Alternatively, the configuration information may be included in the first cryptographic certificate described in relation to Figure 6(a), and the configuration information sent at the same time as the first cryptographic certificate Cblob The second transceiver 15 in the second security context 7 is configured to send the configuration information and configuration certificate C_2Vto the first transceiver 13 in the first security context 5.

In step S214, the signed configuration certificate is verified at the first security context 5. The first processor 17 in the first security context 5 is configured to verily the signed configuration certificate C_2V The signed configuration certificate C₂v is verified using the public half of the second identity key Ka_D pub stored in the device memory 9,

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 after being received and processed in steps S204 to S206. The first processor 17 is configured to perform a signature verification algorithm that, given the signed configuration certificate C_2V and the public key K₂|_{D pub} either accepts or rejects the message's claim to authenticity. This allows the first security context 5 to prove that the configuration information has been generated by a known appliance.

In step S215, the first security context validates that the configuration information meets its requirements.

In an embodiment, the configuration information is requested and checked immediately before transfer of Ktenant to the second security context 7. This ensures that up to date configuration information is validated.

In an embodiment, configuration information and a configuration certificate are also generated and sent by the first security context 5 in a similar manner. The configuration information may comprise information relating to the specific configuration applied by the administrator. The configuration information is signed by the private half of the identity key of the first security context 5 and sent to the second security context 7, which validates that the configuration of the first security context 5 meets its requirements.

In the above described method, the tenant registers itself with the service provider. Registration proceeds in the form of mutual authentication of the two parties over an insecure medium. Mutual authentication of the first security context 5 and the second security context 7 is performed using the unique asymmetric identity keys. The public halves of the identity keys are exchanged, together with their matching generation certificates. Each security context validates the generation certificate of the other by checking that the certificate contains the hash of the public half of the identity key and that the certificate has genuinely been created by a trusted manufacturer. The generation certificate of the identity key may be referred to as a warrant Using the established shared root of trust the first security context 5 and second security context 7 establish a secure, authenticated network communication with a verifiably trustworthy partner. The first security context 5 and the second security context 7 each cryptographically validates that the other was built by a trusted manufacturer

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

The step S205 of verifying the signed generation certificate {C21D } «man p*v at the first security context using Kman pub verifies that the second security context 7 is of a known trustworthy origin.

Figure 3 is a schematic illustration of the tenant system 1 and service provider system 3 after trust has been established, and after the relevant keys have been generated in each security context.

The first device memory 9 in the first security context 5 also stores the public half of the identity key of the second security context Kzid pub- Alternatively, this key may be stored outside of the first security context 5 in an integrity protected manner. The first device memory 9 also stores the tenant key K_tonant, generated in step S601 described below, and the second key pair Κ^η^η· generated in step S401 described below.

The second device memory 11 in the second security context 7 also stores the public half of the identity key of the first security context Ki© _pUb· Alternatively, this key may be stored outside of the first security context 5 in an integrity protected manner. The second device memory 1 also stores the first cryptographic key pair and a first cryptographic certificate generated in step S602 described below.

Figure 4(a) is a flow chart showing a method of transferring a signing key, Ktenam-sign from the first security context 5 to the second security context 7, which is part of a method of cryptographic key transfer according to an embodiment of the present invention. In an embodiment, the method of transferring the signing key Ktenam-sign is performed before the cryptographic key K_tenant is sent to the second security context 7. The method of transferring the signing key K_tenam-sign may be performed after the cryptographic key Ktenant is generated.

In step S401, the asymmetric cryptographic key pair K_tenant__signpub and K_tcnant__signprivand corresponding certificate Ctenam-sign are generated in the first security context 5. The asymmetric cryptographic key pair K_tenant__signpub and K_tenant__signpriv are referred to as the second public key and the second private key and the corresponding certificate Ctenant-sign is referred to as the second cryptographic certificate. The first processor 17 in the first security context 5 is configured to generate the asymmetric cryptographic key pair Ktenant-signpub and K_tenant__si and corresponding certificate C_tenanl._S)gn. The

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 second cryptographic certificate, Ctenant-sign may comprise the signed hash of the second public key K_tenant__slgnpub.

In step S402, the second cryptographic certificate, Ct_enant.si_gn is cryptographically signed with the private half of the identity key of the first security context, Kn_D p^. The first processor 17 is configured to cryptographically sign the second cryptographic certificate, Ctenant-sign with the private half of the identity key of the first security context, Ki ID privThe second cryptographic certificate Ctenant-sign thus comprises information from which the origin of the second public key K_tenant__signpub can be validated. The information from which the origin of the second public key K_tenant-_Sign_pub can be validated comprises the signed hash of the second public key K_tenant__signpub. The second cryptographic certificate Ctenant-sign comprises a hash of the second public key Ktenant-sigitpub ^and'^{s s}'9^necl with the private half of the identity key of the first security context, id _priv, which allows the origin of the second public key K_tenant__slgnpub to be validated.

In step S403, the second public key K_tenant-_Sign_pub and the second cryptographic certificate {Ctenant-sign} KiiD priv ^are sent to the second security context 7. The first transceiver 13 is configured to send the second public key K_Cenant__signpub and the second cryptographic certificate {Ctenam-sign} kuc _ρλ, to the second security context 7.

In step S404, the second cryptographic certificate {Ctenant-sign} khd priv is verified at the second security context 7. The second processor 19 in the second security context 7 is configured to verify the second certificate {Ctenant-sign} kiid priv· The second certificate is verified using the public half of the identity key of the first security context, Kuo pub The second processor 19 is configured to perform a signature verification algorithm that, given the signed message {Ctenant-sign} khd priv and the public key Kn_DpUb either accepts or rejects the message's claim to authenticity.

The authenticity of the second public key K_tenant__signpub is then validated at the second security context 7. The second processor 19 in the second security context 7 is configured to validate the second public key K_tenant__sign In an embodiment in which

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 the second cryptographic certificate Ctenam-sign comprises a hash of the second public key K_tenant__stg_npub, the authenticity of the second public key K_tenant__signpub is validated by calculating the hash the second public key K_tenant-signp_Ut,. end validating that it matches that contained in the second cryptographic certificate C_tenani-sign

If the signature is verified and the second public key validated, the second public key ^Ktenant-signp_Ub> ^is stored in the second device memory 11 of the second security context 7. Alternatively, it can be integrity protected by the second security context 7 and stored in untrusted storage outside of the second security context 7.

If the signature is not verified or the second public key not validated, an error is returned to the first security context 5, for example a message is sent stating “AccessDenied”. At this point, the communication between the first security context 5 and the second security context 7 is terminated.

In the above described method, upon successful establishment of trust, the first security context 5 generates an asymmetric key, Ktenant-sion. and forwards the public half including a certificate signed by the private half of K_1C to the second security context 7. The second security context 7 validates the certificate and stores the public half of Kenants^n in the second device memory 11 or elsewhere in a secure, tamper-proof format for later use. Thus the first security context 5 in the tenant system generates an asymmetric key, Ktenam-sign, and forwards the public half to the second security context for storage and later use.

Figure 4(b) is a schematic illustration of a method of tenant registration, which is part of a method of cryptographic key transfer according to an embodiment of the present invention. The method comprises generating the tenant cryptographic key Kt_enant as described in relation to step S601 below, establishing trust as described in relation to Figure 2(a) above, and generating and exchanging the second public key as described in relation to Figure 4(a) above.

Each vertical box in the diagram represents the enclosed entity, i.e. the first security context 5, second security context 7 and time source 3 over time, with time increasing in the downward direction. The blocks which arrows and loops originate from and terminate at indicate the duration of a particular process. For example, the second

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 security context 7 receives the public half of the identity key of the first security context Kuo pub and corresponding generation certificate. This starts a process in the second security context 7. The next step of the process is to validate the certificate and then forward a response to the first security context 5, i.e. an error message if the certificate is not validated, or a message containing its own public identity key and certificate if the certificate is validated. This is the end of the particular process.

Loops indicate actions that happen internally to a process, for example, validation, which does not require interaction with any other entity. Lines that cross between entities indicate communication between the entities.

The cryptographic key Kt_enant is generated in the first security context 5. The second cryptographic key Ktenam-sign is then generated in the first security context 5. The public half ofthe identity key K_1CpUb and signed certificate {Cw} _Kmanp<iv are then sent from the first security context 5 to the second security context 7 and are validated at the second security context 7. The public half ofthe identity key K₂|_Dpub and signed certificate {C_2D} Kman priv are then sent from the second security context 7 to the first security context 5 and are validated at the first security context 5. The second cryptographic key Ktenant-sign put, and signed certificate {C_tenant-sign} kud pdv are sent to the second security context 7 where they are validated and stored.

Figure 5 is a schematic illustration of the tenant system 1 and service provider system 3 after the second public key, K_lenam-sign pub, has been exchanged during the data transfer process discussed in relation to Figure 6(a) below.

In an embodiment, Kbiob and 0«<λ are transient, and once Ktenam has been transferred, are deleted from the second security context 7.

The second device memory 11 in the second security context 7 also stores the public half of the second key Ktenam-sion pub· Alternatively, this key may be stored outside of the first security context 5 in an integrity protected manner.

Figure 6(a) is a flow chart showing a method of transferring data from the first security context 5 to the second security context 7 in accordance with an embodiment of the

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 present invention. In the embodiment shown in Figure 6(a), the data is a cryptographic key, Ktenant

Once the tenant system has registered itself with the service provider system, the first security context 5 connects to the second security context 7 via the secured, authenticated connection and initiates the key transfer, or import, process.

In step S601, the cryptographic key, Kt_enant> and a corresponding access control list, ACL, are generated in the first security context 5. The first processor 17 in the first security context 5 is configured to generate the cryptographic key, K_lenan(, and corresponding access control list.

The first security context 5 thus generates a key, K_tenant, which they will be lending to the service provider. Ktenant is generated inside a secure appliance in the first security context 5 with an access control list (ACL).

The ACL specifies that a valid use credential, for example a use certificate, must be presented in order to grant a first type of use of the key. The first type of use may be a cryptographic operation for example. The ACL describes how the key may be used, any restrictions on its use and what credentials must be supplied for each operation to be enabled.

In an alternative embodiment, the cryptographic key, Ktenant, is not generated in the first security context 5, but is generated outside the first security context 5 and then provided to the first security context 5.

In an alternative embodiment, instead of the cryptographic key, K_leflanI, some other kind of data is generated at or provided to the first security context 5. An access control list corresponding to the data is then generated, which specifies that a valid use credential must be presented in order to grant a first type of use of the data. The first type of use may be reading the contents of the data file for example.

In the below description, the methods and apparatus are described with reference to the transfer and lease of a tenant key, however it is understood that some other kind of

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 data could be substituted for the tenant key, and transferred and leased in the same manner.

The ACL may comprise one or more permissions, or policies. Each permission may govern a particular use of the key. For example, a first permission governs how the key may be stored, a second permission governs how the key may be used for encryption, a third permission governs how the key may be used for decryption and so on.

The ACL also comprises credential(s) associated with a particular permission. The credentials specify what must be provided to authorize a particular permission. Some permissions may not have an associated credential, for example the storage permission.

The tenant has validated that the second security context 7 adheres to the requirements in the ACL before the key is transferred, as described in relation to Figures 2(a) and (b). The tenant validates that the second security context 7 is manufactured by a trusted manufacturer. The tenant may also validate that the current configuration of the second security context 7 meets the security requirements of the tenant. By validating that the second security context 7 is trusted and meets the tenants requirements, the tenant can be sure that the second security context 7 will enforce the policies/permissions contained in the ACL.

The access control list specifies that a valid use credential, for example a use certificate, must be presented in order to allow certain use(s) of the cryptographic key, Ktenam. Thus the Ktenant ACL may require a valid certificate to be presented each time the key is used for encryption, for example. The permissions related to these uses thus have associated credentials, for example the use certificate.

A tenant may specify that a valid use certificate is required in the permission that governs using a key for encryption for example. Thus in order to use the cryptographic key, Ktenant for encryption, a valid use certificate would need to be presented to activate the permission.

The permission may specify that the use certificate must comprise information from which the cryptographic key Ktenam corresponding to the use certificate can be identified

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 in order to grant the use corresponding to the permission. The ACL may comprise the hash Of Ktenant·

The ACL may comprise a reference to the second security context 7. The reference to the second security context 7 may comprise the hash of the public half of the identity key Kao pubIn an embodiment, the permission requires presentation of a certificate signed by the private half of an asymmetric key, Ktenant-sign, owned by the tenant in order to grant the use associated with the permission. The permission specifies that a use certificate must be signed by the second private key Ktenant-signprj_V 'ⁿ order to grant the use of the cryptographic key, KtenantIn an embodiment, the permission specifies that the use certificate must comprise information from which the expiry of the use certificate can be determined and must be unexpired in order to grant the use of the cryptographic key, Kenam associated with the permission. The tenant is thus able to specify, in the use certificate, an expiry time for its key, after which the key may not be used until further authorization is provided. This allows transfer of cryptographic material between tenant and service provider with assurances over when a key may be used. It ensures that the use of the key is only possible for a set amount of time as specified by the tenant

In an embodiment, the ACL comprises a permission which specifies that foe use certificate must comprise information from which the “start time” of a validity period of foe use certificate can be determined, and the start time must have passed in order to grant use of the cryptographic key, Kt_enant associated with the permission. The tenant is thus able to specify a validity period for the key, outside of which the key may not be used until further authorization is provided. It ensures that the use of the key is only possible for a set time interval, as specified by foe tenant

For the permission that governs storage of the cryptographic key, Kt_enant. the certificate credential may not be present. This means the permission governing storage is always active, or only active until the key is stored in a non-volatile medium, for example. This allows the second security context 7 to store the key without a certificate. For the case in which the permission governing storage is only active until the key is stored in a nonDate Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 volatile medium, the permission is a temporary permission and deactivates after storage, i.e. it’s stripped from the ACL.

The permission may specify that the cryptographic key, Kt_enant can be stored outside the second security context 7 only when encrypted for storage by a key which cannot leave the second security context 7. The permission may specify that Kt_enant can only be encrypted with a key that cannot leave the second security context 7 and is not controllable by the administrator, i.e. the service provider. The permission may also include the key type(s) and mechanism(s) that can be used to encrypt the tenant key Ktenant. for example, the tenant key Ktenant can only be encrypted using AES-GCM encryption with an AES key of 256 bits. The permission may specify that the cryptographic key, K^m can also be stored inside the second security context 7.

In an embodiment, the permission specifies that the cryptographic key K_tenanl and ACL are to be encrypted with an authenticated encryption algorithm.

In an embodiment, the permission specifies that information from which the origin of the cryptographic key K_tenant can be identified is stored in the same data structure as the encrypted cryptographic key Ktenant and access control list, and is authenticity protected.

Alternatively, the permission may specify that the cryptographic key, K_ienant can only be stored inside the second security context 7.

The Ktenant ACL contains a permission which specifies that the tenant key can only be stored inside the second security context 7 and/or outside the second security context 7 if encrypted for storage outside the second security context 7 by a key which is not obtainable by the service provider. As long as the second security context 7 enforces the policy specified in the ACL, the tenant key Ktenant is not exposed to the hosting service provider system 3.

The tenant has ensured that the second security context 7 will enforce the policy because it has received information validating that the manufacturer of the second security context 7 is a trusted manufacturer. Information identifying the manufacturer of the second security context is sent in the generation certificate. For example, the

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 generation certificate is signed with the manufacturer private key. Furthermore, the tenant has received information validating the product, software and hardware of the second security context 7. This information is sent in the generation certificate. As the manufacturer has been identified as a trusted manufacturer, the tenant trusts that the second security context 7 conforms to the information provided by the manufacturer. For example, if the generation certificate contains information specifying that the software and hardware is impervious to attack by the service provider, the tenant trusts this is the case, as the information is provided by the trusted manufacturer. The tenant can validate from the generation certificate that the service provider is using a second security context 7 having certain guarantees from the manufacturer. The manufacturer can provide information in the generation certificate which guarantees that an ACL will be enforced by the second security context 7. As the tenant trusts the manufacturer, this guarantee is also trusted. Thus the guarantee information that allows the tenant to ensure that the ACL will be enforced originates from the manufacturer. Furthermore, the configuration information sent in the configuration certificate informs the tenant whether someone has tried to tamper with the second security context 7 either physically or by attempting to hack it. Once the tenant confirms that the second security context 7 has not been tampered with, then the guarantees from the manufacturer are still applicable.

The ACL allows transfer of cryptographic material between tenant and service provider, with no exposure of the raw cryptographic material to the service provider or to a security agency with jurisdiction over the service provider. For example, a tenant based in the UK would not want their cryptographic keys accessible by an American service provider for fear of the keys being turned over to an American security agency which does not have jurisdiction over a UK-based company.

Furthermore, it allows the service provider to support keys from multiple tenants inside a single infrastructure, removing the need to maintain a dedicated appliance per tenant The tenant would not want their keys accessible to a third party, as a third-party which has been able to extract a tenant's key could then use the key. Providing the access control list comprising a permission which specifies that the cryptographic key, K_tenant can only be stored inside the second security context 7 and/or outside the second security context 7 if encrypted for storage by a key which cannot leave the second security context 7 allows multiple tenants to use the same cryptographic infrastructure

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 whilst ensuring that the raw key material is never exposed to the other tenants. Service providers and tenants are able to share cryptographic infrastructure in a secure way to improve efficiency.

A permission or permissions contained in the ACL thus restrict how Ktenant can be stored and are set so that K_tenant is inaccessible by anyone including the service provider. As a permission corresponding to a first type of use of the key can only be activated by a certificate credential signed by a private key known only to the tenant, the second security context 7 will disallow the first type of use of the key until such a time as an authorizing certificate is presented with the key. This means that the first type of use of the raw key material is not accessible by any third-party including the service provider. The tenant is thus able to maintain assurance that their cryptographic keys are not exposed to either the service provider or any other party. The tenant is able to control the use of its key once imported into the second security context via the access control list.

The access control list may specify that particular information must be provided in the use certificate in order to allow particular uses of the tenant key K_tenant. In this way, the tenant can restrict how the key is used by the service provider. For example, the ACL may contain a permission that specifies that the tenant key can be used to encrypt data when a certificate signed with a certain private key is presented. Thus the credential associated with the permission is the certificate signed with the certain private key. It may also include a permission that specifies decrypt is available when presented by a certificate signed by a different private key. Each permission may require a different credential, for example a certificate signed by a different private key.

Further information that a permission may specify is contained in the certificate may include the time the certificate was issued, where the time stamp comprises a reference to a shared trusted clock and/or the identity (for example the hash of K|_Dpub) of the appliance the certificate can be used with. Including the identity restricts which appliances within a context a key can be used.

The established trust with the second security context and the provision of the ACL corresponding to the tenant key Kt_enani allows the tenant to cryptographically validate that a service provider adheres to a set of rules, i.e. those provided in the ACL, which

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 guarantee the non-disclosure of raw cryptographic material and usage once a key has been transferred. It allows the tenant to securely and verifiably lease cryptographic keys to the service provider for use in a restricted manner, whilst supporting multitenancy.

The ACL comprises policies which ensure that the service provider has no access to the raw key material of a tenant, the tenant has assurance that their key is only usable in situations under its control, and multiple tenants can share the same cryptographic infrastructure in a secure manner. It allows lease of the key Kt_enant whilst restricting the time period in which it may be used. It allows tenants to accurately control the use of their keys once they're hosted by a service provider.

After K_tenant is generated, the tenant then initiates a key-lending protocol with the service provider. For example, the first security context 5 may send a message to the second security context 7 which initiates step S602. The message may request Kblob pub.

In step S602, a first cryptographic key pair which is suitable for use in an encryption scheme and a first cryptographic certificate Cblob are generated in the second security context 7, the first cryptographic key pair comprising a first public key, K_Blob pub, and a first private key, Kblob p<iv In an embodiment, the first cryptographic certificate comprises a hash of the first public key, Kblob pubIn an embodiment, a new first cryptographic key pair K_BLob and first cryptographic certificate Cblob are generated for each data transfer, in other words, each first cryptographic key pair Kblob and first cryptographic certificate Cblob is only valid for a single use.

In an embodiment, the first cryptographic certificate Cblob and the configuration certificate are a single certificate. In other words, instead of the method of Figure 2(b), the first cryptographic certificate is generated further comprising information relating to the current configuration of the second security context. The information relating to the current configuration of the second security context may be as described in relation to Figure 2(b).

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

In step S603, the first cryptographic certificate Cblob is cryptographically signed with the private half of the identity key of the second security context, K_{2E Ρ}πν· The second processor 19 is configured to cryptographically sign the first cryptographic certificate Cblob with the private half of the identity key of the second security context,

In an embodiment, the first cryptographic certificate Cblob validates that Kblob pub was generated in the second security context 7. The first cryptographic certificate comprises information from which the origin of the cryptographic key Kblob pub can be validated. In an embodiment, the information from which the origin of the cryptographic key Kblob pub can be validated is the signed hash of the first public key, Kblob pub. In other words, the first cryptographic certificate Cblob comprises a hash of the first public key Kb_LOb pub and is signed with the private half of the identity key of the second security context, K_aDwhich allows the origin of the first public key, Kblob pub to be validated.

The first cryptographic certificate is signed by K₂id priv in order to prove that the service provider generated the certificate. Including the hash also validates that the data, i.e. the first public key, Kblob pub has not been tampered with during transit

In an embodiment, the first cryptographic certificate comprises information which validates that Kblob _Pdv is ephemeral and that Kblob cannot leave the second security context Information indicating that Kblob pm is ephemeral and that Kblob pnv cannot leave the second security context is sent with the first cryptographic key pair Kblob to the first security context 5. The first cryptographic certificate comprises this information and is signed, thus validating the information contained in the message.

In step S604, the first public key, K_BLOb _pUb, and the signed first cryptographic certificate {Cblob}k2id pnv are sent to the first security context 5. Information relating to Kblob pub, for example indicating that Kblob priv is ephemeral and that Kblob priv cannot leave the second security , may also be sent with the first cryptographic key pair Kblob to the first security context 5 and be included in the first certificate. The second transceiver 15 is configured to send the first public key, Kblob put and the signed first cryptographic certificate {CblobIkzid priv to the first security context 5. Configuration information and/or information about the key may be sent to the first security context 5 in the same message.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

Thus in steps S602 to S604, the second security context 7, in response to the initiation message, generates an asymmetric key, K_Blob> and forwards the public half and a certificate signed by the private half of the identity key of the second security context, K21D p™ to the first security context 5. Kblob is an ephemeral asymmetric key, generated inside the second security context 7 with an accompanying key certificate, C_Blob· Cblob allows the tenant to validate that the key was generated by an appliance inside the second security context 7, that the private half of Ki** is ephemeral and can never leave the second security context 7. The service provider sends the public half of K_BLob and C_Blob to the first security context 5.

In step S605, the signed first cryptographic certificate {C_Blob}k2id pdv is verified at the first security context 5. The first processor 17 in the first security context 5 is configured to verify the signed first cryptographic certificate {C_BLob}k2id puv- The signed first cryptographic certificate {C_BLob}k2id pdv is verified using the public half of the identity key of the second security context, Kao pub- The first processor 17 is configured to perform a signature verification algorithm that, given the signed message {Cblob} k2d puv and the public key K2_{D pub} either accepts or rejects the message's claim to authenticity.

The authenticity of the first public key, Kblob _pUb is then validated at the first security context 5 from the first cryptographic certificate Cblob The first processor 17 in the first security context 5 is configured to validate the first public key, Kblob pub- In an embodiment in which the first cryptographic certificate Cblob comprises a hash of the first public key, Kblob pub. the authenticity of the first public key, K_Blob pub is validated by calculating the hash of the first public key, Kblob pub, and validating that it matches that contained in the first cryptographic certificate Cblob· Any other information sent with the first public key, K_BLob pub, for example current configuration information and information relating to the first public key, Kblob pub, is also validated from the certificate.

If the signature is verified, the first public key, Kblob pub, is stored in the first device memory 9 of the first security context 5. Alternatively, it can be integrity protected by the first security context 5 and stored in untrusted storage outside of the first security context 5.

If the signature is not verified or the key is not validated, an error is returned to the second security context 7, for example a message is sent stating “AccessDenied. At

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 this point, the communication between the first security context 5 and the second security context 1 is terminated.

In an embodiment in which the first cryptographic certificate Cblob is generated further comprising information relating to the current configuration of the second security context, the information relating to the current configuration of the second security context is also validated in step S605, as described in relation to Figure 2(b). Any key information included in the message is also validated.

Thus in step S605, the first security context may use the certificate to validate the parameters of Kblob, authenticating the source and checking that it is well-formed and conforming to the security policies expected by the tenant. The security policy a tenant may require could include a key length, key type and/or key permission-credential pairs. Key permissions are used to describe the operations that a key may be used for. For example, permissions may specify that a key can be used to encrypt other keys or that it may be used for cryptographically signing data. Permissions must be activated by a matching credential before they can be used. A credential in this system takes the form of a verifiable cryptographic certificate. The first security context 5 validates Cblob. that the private half of Kblob is ephemeral, that the second security context 7 is manufactured by a trusted source, and that the state of the second security context 7 is suitable for lending a key to.

Step S606 comprises encrypting the cryptographic key K_tenani and the corresponding access control list with the first public key Kblob pub in the first security context 5. The first processor 17 is configured to encrypt the cryptographic key Ktenant and the corresponding access control list with the first public key Kblob pub·

As the communication channel between the first security context 5 and the second security context 7 is provided and controlled by the service provider, it is not trusted by the tenant for transfer of high value cryptographic keys such as Ktenant, as it could be open to attack by the service provider. Thus the security of Kenant is enforced by the tenant by encryption of Ktenant with the first public key Kblob pub· The first certificate Cblob allows the tenant to validate that the private half of K^b can never leave the second security context 7, thus by encryption of Ktenant with the first public key Kblob pub. the

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 tenant can ensure that K^nt is secure from the service provider, even though the service provider can attack the communication channel.

The encryption of the tenant key in this way provides a secure channel which is rooted in the trust between the first security context 5 and the second security context 7. This allows a higher level secure authenticated channel that is not trusted by the first security context 5 and the second security context 7 to be used between the tenant and service provider site. This higher layer secure channel may be provided by the service provider using a load balancer or firewall appliance. The encryption of the tenant key allows the service provider to continue to use this infrastructure to mitigate attacks such as denial of service but still allows context to context security without having to trust external security services.

Kblob pub may be discarded by the tenant after encryption of tenant.

In an embodiment, the information from which the origin of the cryptographic key K_ten_8nt can be validated is the signed encrypted tenant key and access control list {{Ktenant. ACL} KBLOB pub} Ktenant-sign ρήν·

Thus in step S607, the output blob from step S606 is cryptographically signed using the private half of Ktenant-sqn· The first processor 17 is configured to cryptographically sign the output blob from step S606 with the private half of Ktenant-sign·

In step S608, the encrypted cryptographic key and corresponding access control list {Ktenant, ACL} kblob pub, and information from which the origin of the cryptographic key Ktenant can be validated is sent to the second security context 7. In an embodiment information from which the origin of the message can be identified is also sent, which may be the hash of the public half of Ktenant-sign for example

In an embodiment, in which the information from which the origin of the message can be identified comprises a hash of Kte_nant-sign_pub, the encrypted blob, hash of the signing key and signature are forwarded to the second security context 7. The signature is information from which the origin of the cryptographic key Ktenant can be validated. The first transceiver 13 is configured to send the encrypted blob, signed

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 output blob from step S607 and the hash of the public half of K_tenant^gn to the second security context 7.

Given an acceptable key K_BLob pub. that is a key Kblob pub that the tenant defines as being of sufficient cryptographic strength, the key which is to be registered, K_tenant, is encrypted with the public half of K_biob at the first security context 5. As described above, Ktenant includes permission-credential pairs, i.e. the ACL, which specify the acceptable operations for which the key material may be used and which may be activated by a certificate which must be signed by the private half of K^am-sign. The encrypted blob is signed using the private half of Ktenam-sign and the encrypted blob, signature and hash of the public half of Kt_ena_nt._Sign are forwarded to the second security context 7.

The transfer of the cryptographic key K^nant to the second security context in the service provider is secure from the service provider and other attackers, as the key used to encrypt the cryptographic key K^nt is not recoverable in plaintext. is required to successfully decrypt the data sent from the first security context 5 to the second security context 7. Kbiobppv is ephemeral and cannot be accessed or mutated by the service provider or a third party.

Figure 6(b) is a flow chart showing further steps of a method of transferring a cryptographic key, K_tenant from the first security context 5 to the second security context 7 in accordance with an embodiment of the present invention.

In step S609, the origin of the cryptographic key Ktenam is validated at the second security context 7. In an embodiment, this is validated by first verifying the signature contained in the message sent from the first security context 5.

In an embodiment, the hash of the public half of Ktenam-s^n is used to identify the tenant from which the message originates, and thus identify the correct signing key required to verify the signature.

The signature is then verified using the public half of K_tenant-sign· The second processor 19 in the second security context 7 is configured to verity the signature. The signature is verified using the public half of Kt_enant._Sign· The second processor 19 is configured to

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 perform a signature verification algorithm that, given the signature and the public key Ktenant-sgn either accepts or rejects the message's claim to authenticity.

It is then validated that the contents of the signature match the encrypted blob. This allows validation of the origin of the encrypted blob.

If verified and validated, the method progresses to step S610. If not verified or validated, an error is returned to the first security context 5, for example a message is sent stating “AccessDenied”. At this point, the communication between the first security context 5 and the second security context 7 is terminated.

Step S610 comprises decrypting the encrypted cryptographic key K_tenant and corresponding access control list with the first private key K_BLob priv at the second security context 7. The second processor 19 is configured to decrypt the encrypted cryptographic key K_tenant and corresponding access control list. The second processor 19 is configured to perform a decryption algorithm given the encrypted cryptographic key Ktenant and corresponding access control list and the first private key Kblob pitvStep S611 comprises re-encrypting the cryptographic key Ktenant and the ACL with a third cryptographic key at the second security context 1. This step is performed so that the cryptographic key Ktenant and the ACL may be stored outside of the second security context 7. In an alternative embodiment, this step is omitted and the cryptographic key Ktenant, the ACL and information identifying the origin of the cryptographic key Kt_enant are stored unencrypted within the second security context 7.

The ACL specifies the parameters of which type(s) of key can be used to encrypt Ktenant. and which encryption mechanisms can be used.

The ACL may comprise a permission, i.e. a policy, which states that Kt_enant can only be encrypted with a key that cannot leave the second security context 7 and is not controllable by the administrator, i.e. the service provider. The permission also includes the key type(s) and mechanism(s) that can be used to encrypt the tenant key Kt_enant, for example, the tenant key K_!enant can only be encrypted using AES-GCM (Galois/Counter Mode) encryption with an AES key of 256 bits.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

The encryption mechanism may specify the need to ensure authenticity, integrity and confidentiality or some subset of those properties.

The second security context 7 may either re-use an existing key that meets the ACL policy requirements, or create a new key.

The cryptographic key Ktenant and the ACL are encrypted in the second security context 7 using the mechanism specified, and a third cryptographic key which meets the specified requirements. The second processor is configured to encrypt the cryptographic key Ki_enant with a third cryptographic key. In an embodiment, the cryptographic key, Ktenant is encrypted for storage by a key which cannot leave the second security context 7.

The cryptographic key Ktenant and the ACL may be encrypted using an authenticated encryption algorithm.

In an embodiment, the cryptographic key Ktenant and ACL are inputs to be encrypted in an authenticated encryption algorithm, for example an AES-GCM operation, whilst information from which the origin of the cryptographic key Ktenant can be identified is input as an authenticated data parameter. In an embodiment, the information from which the origin of the cryptographic key Ktenant can be identified is the public half of Ktenant-sign· In an embodiment a hash of Ktenant is also stored with the cryptographic key Ktenant, for example the hash of k_tenant may be the name ofthe file in which the encrypted data is stored.

The re-encrypted cryptographic key Ktenant and corresponding access control list, and the authenticated information from which the origin of the cryptographic key Ktenant can be identified, are transferred to a device memory outside of the second security context 7.

The information from which the origin of the cryptographic key Kt_enant can be identified is thus stored in the same data structure as the encrypted cryptographic key K_tenant and access control list, but not necessarily be encrypted. However, the information from which the origin of the cryptographic key K_tenant can be identified is authenticity protected, such that it cannot be changed without notice.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

In the above described steps, upon receipt, the second security context 7 verifies the payload signature using the public half of Ktenant-sign. The hash contained in the received payload is used to identify the public half of the tenant's signing key. The received payload signature is verified using the identified signing key. Upon successful validation of the payload, K_tenant is decrypted using the private half of Kbiob and reencrypted under a non-recoverable key, as enforced by the K_tenant ACL, and stored alongside the public half of K_lenant-sign outside of the second security context 7 for later use. Alternatively, the key K^ant is stored unencrypted in protected non-volatile memory within the second security context 7. In both cases, the decrypted key is thus stored in a secure, tamper-proof manner.

The manner in which K_teoani is stored is described by its permissions, or policies. Specifically, at generation time, the permissions granted restrict how Kt_enant can be stored and are set so that Ktenant is inaccessible by anyone including the service provider, other than the second security context 7.

In an embodiment in which a plurality of tenants store a key with a single service provider, each of the tenant keys can be encrypted for storage with the same key. Alternatively, each tenant key may be encrypted with a different key if there are different storage permissions between the ACLs corresponding to each tenant key for example.

Figure 7(a) is a schematic illustration of a method of key enrolment, which is part of a method of cryptographic key transfer according to an embodiment of the present invention. The method comprises steps S602 to S611 described in relation to Figures 6(a) and (b) above.

A key enrolment message is sent from the first security context 5 to the second security context 7. In response to this message, the first cryptographic key pair Κ^οβ is generated at the second security context 7. The second security context 7 sends the public half of the first cryptographic key pair Kblob and the signed first cryptographic certificate {C_Blob) kzid priv to the first security context 5, where they are validated. The cryptographic key Ktenant is encrypted at the first security context 5 with Kblob pub. The encrypted key {Ktenant} kblob pub and hash{Ktenant-sign pub} are signed with Ktenant-sign priv at the

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 first security context 5. The signed message is sent to the second security context 7 where it is validated, and the encrypted key {Kt_enant} kbiob pub is decrypted and reencrypted in the second security context 7 with a non-recoverable key that is safe from the administrator. The encrypted tenant key can then be stored outside of the second security context 7.

Figure 7(b) is a flow chart of a method of cryptographic key transfer according to an embodiment of the present invention.

The tenant sets the ACL of the tenant key in the first security context 5. The first security context 5 then requests the first public key K^qb _pub from the second security context 7. The first security context 5 sends a request message to the second security context 7 for example. The first security context 5 then validates the received first public key Kblob pub and first cryptographic certificate Cblob- This corresponds to steps S605 described above. If the first public key Kblob pub and first cryptographic certificate Cblob are not validated, an error is returned. If they are validated, the first security context 5 encrypts the tenant key and associated ACL with the first public key Kblob pubThis corresponds to step S606 described above. The first security context 5 then signs the encrypted key and associated ACL and a hash of K_leoant.sien pub with K_lenant._sig_npriv. This corresponds to the S607 described above. The first security context 5 then exchanges the encrypted blob, the hash of Kt_enant-sign pub and the signature with the second security context 7. This corresponds to step S608 above.

Figure 8(a) is a schematic illustration of the tenant system 1 and service provider system 3 after the tenant key Ktenant has been imported to the second security context 7. In an embodiment, K^b and C_biob are transient, and once Ktenant has been transferred are deleted from the second security context 7 and the first security context 5.

The first device memory 9 in the first security context 5 also stores the first public key KeLoepub- Alternatively, this key may be stored outside of the first security context 5 in an integrity protected manner. The second device memory 11 in the second security context 7 also stores the tenant key K_tenanl and the corresponding ACL. Alternatively, the tenant key Κ^^ and ACL may be stored outside the second security context 7, encrypted with a third cryptographic key which cannot leave the second security context 7.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

Figure 8(b) is a schematic illustration of a tamper-proof, secure reference time source 2, which may be hosted by the service provider, the tenant or a separate third-party. If hosted by the service provider, the reference time source may be part of the second security context 7, or may be a different security context.

The time source 2 comprises a third device memory 35. The third device memory 35 is configured to store cryptographic information such as keys, key pairs and certificates. The third device memory 35 may include any form of non-volatile device memory such as flash, optical disks or magnetic hard drives for example. The time source 2 also comprises volatile memory.

The third device memory 35 stores a unique asymmetric identity key K_Tsid. with a corresponding signed generation certificate {C_Tsio}Kman ρ«ν K_Tsid is a signing key used to prove the origin of data and authenticity. The generation certificate Ctsid may describe the public parameters of the key, for example information relating to the type of key, and its length. The generation certificate Ctsid comprises information which authenticates that the identity key Ktsid was generated in the time source 2. For example, the generation certificate C_TS|_D may comprise the signed hash of the public half of Ktsid. The generation certificate Ctsid may also include state information, for example, information relating to a unique identification of the device, information identifying the manufacturer, the device version, hardware version, the software type and the supported features of the model. The generation certificate Ctsid is signed by a manufacturer trusted by both the first security context 5 and the second security context 7. The generation certificate is cryptographically signed with the private half of a manufacturer asymmetric key, Kman pdv· The manufacturer may be a third party who manufactured the security appliance(s) which forms the first security context, the security appliance(s) which forms the second security context and the time source 2.

The time source 2 further comprises a third transceiver 39. The third transceiver 39 is configured to transmit and receive data packets. The data packets may be transmitted from and received at the transceiver 39 via a wireless Internet connection for example.

The time source 2 further comprises a third processor 41. The third processor 41 is configured to perform cryptographic operations, such as generation of cryptographic

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 keys and asymmetric cryptographic key pairs, generation of certificates corresponding to a cryptographic key or asymmetric cryptographic key pair, generation of access control lists corresponding to a cryptographic key, generation of use certificates corresponding to a cryptographic key, encryption of an object with a cryptographic key which is stored in the third device memory 35, decryption of an encrypted object with a cryptographic key which is stored in the third device memory 35, cryptographically signing an object with a cryptographic key which is stored in the third device memory 35, signature verification and validation of an object based on information stored in the third device memory 35.

The time source 2 further comprises a clock 37. The clock 37 may be a monotonic clock, i.e. a counter that increases monotonically over time. The clock may represent time of day to a high degree of accuracy. In an embodiment, the clock is accurate to 1 second or better. For example, the clock may be based on a GPS signal which is accurate to the order of 40 nanoseconds.

The time source 2 may be resistant to tamper, for example by the inclusion of physical security such as a membrane that covers the entire device, and that cannot be removed without destroying the underlying physical hardware, thus making it unusable.

The three services shown in Figure 8(a) and 8(b) comprising: 1) the first security context 5, for example a secure appliance, owned and operated by a tenant; 2) the second security context 7, for example one or more secure appliances owned and operated by the service provider; and 3) a trusted time source 2, which may be owned by the service provider, an external third-party or a tenant, are a set of cooperating services that can all verify that they were built by trusted third-parties with a set of known security properties and adhere to certain same rules. There is an assumption that a tenant trusts the aforementioned third-parties which built the cooperating services but does not trust the service provider who is hosting some of these services. A security context may be one or more security appliances that share a set of cryptographic primitives and form a cluster across which requests may be loadbalanced.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

Each of the first security context 5, second security context 7 and reference time source 2 can be verifiably identified using a cryptographic certificate, which may be generated at the time of manufacture for example. Each is able to securely store their identity in a way that means they cannot be imitated. Identifying the first security context 5, second security context 7 or reference time source 2 allows for its origin to be ascertained.

Furthermore, the configuration and state of each of the first security context 5, second security context 7 and reference time source 2 can be validated in a non-repudiable manner. The origin and state of a service define enough information by which other services can place their trust.

Each component contains a K_D generated in the factory when it was manufactured for example. Each component also contains the key certificate for the K_D which is signed using an asymmetric key known only to the manufacturer. The public half of the manufacturer key can be used as the root of trust to authenticate genuine appliances.

The transfer of cryptographic certificates and cryptographic keys between the reference time source 2 and the second security context 7 described in the following may occur over a secure, authenticated channel between the reference time source 2 and the second security context 7, which is provided and controlled by the service provider. Note that although the channel may be secured from third parties, it is open to attack by the service provider, and is thus untrusted.

The exchange of public keys and certificates described in relation to Figure 9 below allows a trusted relationship between the second security context 7 and the time source 2 to be established.

Figure 9 is a flow chart showing a method of registering a time source 2 with the second security context 7. This can be performed at any point as part of the key transfer method according to an embodiment of the present invention, or as part of the method of controlling use of a cryptographic key according to an embodiment of the present invention. For example, the method of registering a time source 2 may be performed when the administrator of the second security context 7 sets up the appliances in the second security context 7, i.e. when authenticity is performed. This is

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 an infrequent process, and may only need to be performed once. After enrolment of the time source 2, the second security context 7 may then broadcast information relating to its trusted time sources to potential tenants. The tenants therefore know initially which time source(s) the service provider supports, and can make a decision as to whether these time sources are acceptable before enrolment of keys begins, for example by validating the authenticity, state and configuration of the time source(s) as described in relation to Figure 10 below.

If the reference time source 2 is part of the second security context 7, it is not necessary to register the time source 2 with the second security context 7 and these steps are not performed.

A reference time source 2 comprises a device memory 35 and a clock 37 as described in relation to Figure 8(b) above.

In step S901, the public half of the identity key of the time source, K_TSId pub. and the generation certificate {Ctsd } Kman pnv. are sent from the reference time source 2 to the second security context 7 in response to a query message from the second security context 7. A transceiver 39 in the reference time source 2 is configured to send the public half of the identity key of the time source, K_Tsid pub. and the generation certificate {Ctsid} Kman priv, to the second transceiver 15 in the second security context 7. Information relating to the state of the time source device 2 may be sent in the same message. The information relating to the state of the device will also be contained in the generation certificate in this case, in order to allow validation of the state information.

In step S902, the generation certificate {Ctsid} Km» pnv is verified at the second security context 7. The second processor 19 in the second security context 7 is configured to verify the generation certificate {Ctsd } Kman priv- The generation certificate is verified using the public half of the trusted manufacturer key Κ™_η put,. The second processor 19 is configured to perform a signature verification algorithm that, given the signed message {Ctsd } Kman _Pnv and the public key K^n pub either accepts or rejects the message's claim to authenticity.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

The authenticity of the public half of the identity key of the time source, Ktsc pub is then validated at the second security context 7. The second processor 19 in the second security context 7 is configured to validate the public half of the identity key of the time source, Ktsid pub. In an embodiment in which the generation certificate Cts» comprises a hash of the public half of the identity key of the time source, Ktsid pub. the authenticity of the public half of the identity key of the time source, Ktsid pub. is validated by calculating the hash of the public half of the identity key of the time source, Ktsid put, and validating that it matches that contained in its generation certificate CtsidIn step S903, the second security context 7 may also validate any state information contained in the message meets requirements.

If the signature is verified and the state information is validated, the public half of the identity key of the time source, K_Tsd pub. is stored in the second device memory 11 of the second security context 7. Alternatively, it can be integrity protected by the second security context 7 and stored in untrusted storage outside of the second security context 7.

If the signature is not verified, or the key or state information is not validated, an error is returned to the time source 2, for example a message is sent stating “AccessDenied”. At this point, the communication between the time source 2 and the second security context 7 is terminated.

In the above described steps, the second security context 7 establishes trust with the reference time source 2. The second security context 7 thus establishes trust with the time source 2 by validating its origin, state and authenticity via a certificate chain.

Once trust has been established, the second security context 7 requests configuration information from the time source 2 in step S904. The transceiver 15 in the second security context 7 sends a message requesting the information to the time source 2.

In step S905, in response to the request, the time source 2 generates configuration information which includes information relating to the configuration of the time source 2, which may include precision information about the clock and security settings, the cryptographic primitives that have been setup, the version of software that the time

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 source is running and the state of the time source 2, including any anomalies detected. The third processor 41 in the time source 2 is configured to generate this message.

In step S906, the response message is signed by the private half of the identity key of the time source, K_TS© ρπν The third processor 41 in the time source 2 is configured to cryptographically sign the message with the private half of the identity key of the time source, Ktsid priv·

In step S907, the signature and configuration information is sent to the second security context 7. The transceiver 39 in the time source 2 is configured to send the signed message to the second security context 7.

In step S908, the second security context 7 verifies the response signature using public half of the identity key of the time source, K_Tsid pub The second processor 19 in the second security context 5 is configured to verify the signature. The signature is verified using the public half of the identity key of the time source, K_TSd pub- The second processor 19 is configured to perform a signature verification algorithm that, given the signed message and the public key Ktsid pub either accepts or rejects the message's claim to authenticity.

In step S909, a check as to whether the time source configuration is within acceptable tolerances is performed. The configuration information is checked to validate that it meets the requirements of the service provider.

The configuration information can be checked again whenever a time stamp is requested by the second security context 7, by requesting configuration information again.

The state and configuration of all appliances in the system, i.e. the first security context 5, the second security context 7 and the reference time source 2 can be obtained from the state entry present in the appliance. After initialization, each appliance updates this field and generates a new configuration message to allow clients to cryptographically verify the new appliance settings.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

If the signature is verified successfully and the configuration of the time source 2 is within tolerances, the method progresses to step S910. If not, an error is returned to the time source 2, for example a message is sent stating “AccessDenied”. At this point, the communication between the time source 2 and the second security context 7 is terminated.

In step S910, a message is sent to the first security context 5 comprising information identifying the time source 2 and information indicating that the time source 2 is trusted by the second security context 7. The information identifying the time source 2 may comprise a unique identification number for example the hash of the public half of the identity key of the time source 2 or the IP address of the time source 2. The transceiver 15 in the second security context 7 is configured to send a message comprising information identifying the time source 2 and information indicating that the time source is trusted to the first security context 5. Information identifying the time source 2 and information indicating that the time source 2 is trusted is also stored in the device memory 11 of the second security context 7, or outside of the second security context 7 in an integrity protected manner.

In an embodiment, the second security context 7 adds the time source 2 to a list of trusted time source(s). The reference time source's unique identification may be added to the list. The list of trusted time source(s) is then broadcast by the second security context 7. The first security context 5 receives the list of trusted time source(s) from the second security context 7.

Given the received information identifying one or more trusted time sources, the first security context 5 can query, and in the same way as the second security context 7, validate the authenticity, state and configuration of the time source(s). A flow chart illustrating this process is shown in Figure 10. Again, this can be performed at any point as part of the key transfer method according to an embodiment of the present invention, or as part of the method of controlling use of a cryptographic key according to an embodiment of the present invention.

Alternatively, if the reference time source 2 is part of the first security context 5, the below steps will not be performed.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

Furthermore, in a case in which the reference time source 2 is identified first by the first security context 5, the below steps may be performed, and then information regarding the time source 2 sent to the second security context 7, which then validates the time source 2.

The transfer of cryptographic certificates and cryptographic keys between the reference time source 2 and the first security context 5 described in the following may occur over a secure, authenticated channel between the reference time source 2 and the first security context 5. Where the time source is provided by the service provider, the channel may be provided and controlled by the service provider or a proxy. Where the time source is provided by a third party time service provider, the channel may be provided directly between the tenant and time service provider. This channel may be provided by the time service provider or a proxy. Note that although the channel is secure from third parties, it is open to attack by the time service provider. The transfer of cryptographic certificates and cryptographic keys between the reference time source 2 and the second security context 7 described in the following may occur over a secure, authenticated channel between the reference time source 2 and the second security context 7.

The exchange of public keys and certificates described in relation to Figure 10 below allows a trusted relationship between the first security context 5 and the time source 2 to be established.

In step S1001, the public half of the identity key of the time source, K_Tsid pub, the generation certificate {Ctsid } «man _Pnv, and a configuration certificate {Ctsd } Kman pm are sent from the reference time source 2 to the first security context 5 in response to a query message from the first security context 5. A transceiver 39 in the reference time source 2 is configured to send the public half of the identity key of the time source, Ktsid pub, and the generation certificate {C_Tsio} Kman pnv, to the first transceiver 13 in the first security context 5. Information relating to the state of the time source device(s) may be sent in the same message. The information relating to the state of the device(s) will also be contained in the generation certificate in this case, in order to validate the state information.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

In step S1002, the generation certificate {Ctsid } Kman pnv is verified at the first security context 5. The first processor 17 in the first security context 5 is configured to verify the generation certificate {Ctsid } Kman pw The generation certificate is verified using the public half of the trusted manufacturer key Kman pub. The first processor 17 is configured to perform a signature verification algorithm that, given the signed message {Ctsid} Kman pm and the public key K^n _pub either accepts or rejects the message's claim to authenticity.

The authenticity of the public half of the identity key of the time source, Ktsd pub is then validated at the first security context 5. The first processor 17 in the first security context 5 is configured to validate the public half of the identity key of the time source, Ktsid pub- In an embodiment in which the generation certificate C-so comprises a hash of the public half of the identity key of the time source, Kts© put. the authenticity of the public half of the identity key of the time source, K_Tsid pub. is validated by calculating the hash the public half of the identity key of the time source, Ktsid pub. and validating that it matches that contained in its generation certificate C_TSd.

In step S1003, the first security context 5 may also validate that the state of the device(s) meets requirements, from any state information contained in the message.

If the signature is verified and the key and state information is validated, the public half of the identity key of the time source, K_Tsid p«b, is stored in the first device memory 9 of the first security context 5. Alternatively, it can be integrity protected by the first security context 5 and stored in untrusted storage outside of the first security context 5.

If the signature is not verified or the key or state information is not validated, an error is returned to the time source 2, for example a message is sent stating “AccessDenied”. At this point, the communication between the time source 2 and the first security context 5 is terminated.

In the above described steps, the first security context 5 establishes trust with the reference time source 2. The first security context 5 thus establishes trust with the time source 2 by making a secure connection to the time source, and validating its origin and authenticity via a certificate chain.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

Once trust has been established, the first security context 5 requests configuration and state information from the time source 2 in step S1004. The transceiver in the first security context 5 sends a message requesting information to the time source 2.

In step S1005, in response to the request, the time source 2 generates a message which includes the configuration of the time source 2, which may include precision information about the clock and security settings, the cryptographic primitives that have been setup, the version of software that the time source is running and the state of the time source, including any anomalies detected. The third processor 41 in the time source is configured to generate the configuration information.

The message may be the same message sent to the second security context 7. Alternatively, a message containing different information may be generated and sent in a similar manner as described in relation to Figure 2(b). The message may comprise information relating to the specific configuration applied to the reference time source.

In step S1006, the response message is signed by the private half of the identity key of the time source, K_{TSÜ prtv}. The third processor 41 in the time source is configured to cryptographically sign the message with the private half of the identity key of the time source, KysiDprivIn step S1007, the message and signature is sent to the first security context 5. The transceiver 39 in the time source 2 is configured to send the signed message to the first security context 5.

In step S10O8, the first security context 5 verifies the response signature using public half of the identity key of the time source, Ktsid pub· The first processor 17 in the first security context 5 is configured to verify the signature. The signature is verified using the public half of the identity key of the time source, Ktsid pub· The first processor 17 is configured to perform a signature verification algorithm that, given the signed message and the public key K_TS|_D either accepts or rejects the message's claim to authenticity.

In step S1009, a check as to whether the time source configuration and the time source state are within acceptable tolerances is performed. The information is checked to validate that it meets the requirements of the tenant In an embodiment, the

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 configuration information is requested and checked immediately before transfer of Ktenant to the second security context 7. This ensures that up to date configuration information is validated.

If the signature is verified successfully and the configuration and state of the time source 2 are within tolerances, information identifying the time source 2 and information indicating that the time source 2 is trusted are stored in the device memory 9 of the first security context 5, or outside of the first security context 5 in an integrity protected manner. If not, an error is returned to the time source 2, for example a message is sent stating “AccessDenied”. At this point, the communication between the time source 2 and the first security context 5 is terminated.

The configuration information can be checked again whenever a time stamp is requested by the first security context 7, by requesting configuration information again.

Figure 11 is a flow chart showing a method of controlling use of a cryptographic key, Ktenant according to an embodiment of the present invention. The cryptographic key, Ktenam is securely stored in the service provider system together with an access control list specifying that a valid use credential, for example a use certificate, must be presented in order to allow a first type of use of the cryptographic key, K_tenant.

In an alternative embodiment, instead of the cryptographic key, K_tena_nt, some other kind of data is securely stored in the service provider system, and the use of this data is controlled. An access control list is stored together with the data, which specifies that a valid use certificate must be presented in order to grant a first type of use of the data. In the below, the methods and apparatus are described with reference to the use of a tenant key, however it is understood that some other kind of data could be substituted for the tenant key, and used in the same manner.

The process of enabling the use of an imported key can be referred to as Key Authorization.

At the point at which the tenant wishes to make its key available for use in the service provider's cryptographic infrastructure, the tenant generates a use certificate.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

Step S1101 comprises generating a use credential in the first security context 5. In an embodiment, the use credential is a use certificate. The processor 17 in the first security context 5 is configured to generate the use certificate. The use certificate comprises information from which the cryptographic key Kt_enant corresponding to the use certificate can be identified and information from which the expiry of the use certificate can be determined.

In an embodiment, the information from which the cryptographic key Kenan· corresponding to the use certificate can be identified comprises the hash of K_lenant·

In an embodiment, the information from which the expiry of the use certificate can be determined comprises an expiry time, and information identifying a reference time source. The expiry time is calculated with reference to a reference time source 2 that is trusted by both the first security context 5 and the second security context 7. Trust in a reference time source is ascertained by the first security context 5 and the second security context 7 independently connecting to the reference time source and validating that the time source is trusted and that the time source is tamper-proof via the module state parameter of the certificate. This is described in relation to figures 9 and 10 above and may be performed prior to the method of key transfer, or after the method of key transfer and before the generation of a use certificate for example.

The use certificate thus comprises information regarding a validity period with reference to a secure time source 2.

Key Authorization thus begins with a tenant generating a credential certificate, or a “key-use” certificate that matches that defined in a permission of Ktenam. the ACL. In other words, the use certificate fulfils the requirements which are defined in the access control list corresponding to K_tenant.

In an embodiment, the credential certificate, or use certificate, is signed with the private half Of Ktenant•sign*

Figure 12 shews a method of generating a use certificate at the first security context 5, which is part of a method of controlling use of a cryptographic key, K_lenanl according to an embodiment of the present invention.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

In step S1201, a reference time source 2 is selected by the first security context 5. Information identifying one or more time sources and information indicating that the time source(s) are trusted may be stored in the device memory 9 of the first security context 5 or outside of the first security context in an integrity protected manner. One of the time sources is selected by the first security context 5.

The time source 2 may be chosen from a list of trusted time sources broadcast by the second security context 7 and for which the first security context 5 has established trust. Thus both the second security context 7 and the first security context 5 have established trust with the one or more time sources on the list. Each of these time sources are valid time sources with the necessary configuration and state. The first security context selects a time source 2 from the list of one or more valid time sources.

In step S1202, the first security context 5 requests the current time stamp from the chosen time source 2. The first transceiver 13 in the first security context 5 sends a request message to the time source 2.

In step S1203 the time source 2 generates a message including the time stamp. The message may comprise information identifying the time source and the current time at the generation of the message, determined from the time source clock. The information identifying the time source may comprise the hash of the public identity key of the generating time source, K_Tsid pubThe message may further comprise the current configuration information, relating to any indication that the time source has been tampered with for example. This information can then be used by the client to check whether the time source is in an acceptable state before using the timestamp in the use certificate.

In step S1204, the message is signed with the private half of the identity key of the time source, K_TS|_{D P}n_V- The third processor 41 in the time source is configured to cryptographically sign the message with the private half of the identity key of the time source, K_TSiopriv·

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

In step S1205, the message and signature is sent to the first security context 5. The transceiver 39 in the time source 2 is configured to send the signed message to the first security context 5.

In step S1206, the first security context 5 verifies the response signature using the public half of the identity key of the time source, Ktsid pub The first processor 17 in the first security context 5 is configured to verify the signature. The signature is verified using the public half of the identity key of the time source, Ktsd pub· The first processor 17 is configured to perform a signature verification algorithm that, given the signed message and the public key Ktsid pub either accepts or rejects the message's claim to authenticity. If the signature is verified successfully the method progresses to step S1207. If not, an error is returned to the time source 2, for example a message is sent stating “AccessDenied”. At this point, the communication between the time source 2 and the second security context 7 is terminated.

If configuration information is included in the message, the first security context 5 validates that the configuration is acceptable. If the current configuration information is not acceptable to the tenant, then the use certificate generation is stopped.

In step S1207 the tenant calculates the expiry time for their key K_lenant based on the aforementioned time stamp. The expiry time is calculated as the time stamp time plus an amount of time for which the tenant wishes the certificate to be active, for example the time stamp time plus one or more days.

Where the expiry time is small compared to the communication delay, a communication delay may be factored in to the expiry time for example. In this case, the expiry time is calculated as the time stamp time plus the communication delay from the reference time source 2 to the first security context 5 plus the communication delay from the first security context 5 to the second security context 7 plus the amount of time for which the tenant wishes the certificate to be active.

The amount of time for which the tenant wishes the certificate to be active may be a few seconds or several days for example, depending on the application.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

A “start time” may also be included in the use certificate, such that the use certificate defines a validity period, within which the certificate is valid. This allows a tenant to pregenerate one or more certificates, in advance of use.

In step S1208, the expiry time, a reference to the time source, a reference to Kt_enantand a reference to the second security context 7 are included in a use certificate.

A reference to a system component, such as the time source or the second security context 7 may be a unique ID that can be tied to it using a certificate, or could be the hash of the public half of its identity key Ko.

The processor 17 in the first security context 5 is configured to generate a use certificate, comprising the expiry time calculated in step S1207, a reference to a time source, a reference to K_tenantand a reference to the second security context 7.

In an embodiment, the reference to K^am is the hash of K_tenant· In an embodiment, the reference to the time source 2 is the hash of the public half of its identity key Κ_Τ5!ο-ρ<*· In an embodiment, the reference to the second security context 7 is the hash of the public half of its identity key K₂iD.pub·

Returning to Figure 11, once the use certificate has been generated, it is issued together with information from which the origin of the use certificate can be identified. In an embodiment, the information from which the origin of the use certificate can be identified is a hash of K_tenant__sig_npub.

In step S1102, the use certificate is signed with the private half of the signing key Ktenant-sign priv. The first processor 17 in the first security context 5 is configured to cryptographically sign the certificate with the private half of the signing key K_tenaRt._sen _Pnv.

In step S1103, the use certificate is sent to the application server of the service provider, together with the information from which the origin of the use certificate can be validated. The information from which the origin of the use certificate can be validated is the signature, i.e. the signed use certificate.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

The first transceiver 13 is configured to send the use certificate to the application server. Other entities which wish to use the tenant key may be sent a use certificate. These entities may communicate directly with the second security context 7 or via the application server.

Information from which the origin of the use certificate can be identified may also be sent with the use certificate. The information from which the origin of the use certificate can be identified may be a hash of The information from which the origin of the use certificate can be identified is included in order to identify which signing key should be used to verify the signature.

The first security context 5 thus sends the use certificate, the signature and the hash of Ktenant-sign put» to the service provider. Each use of the key by the service provider is then accompanied by the presentation of this information to the second security context 7.

In step S1104, the use certificate, signature and the information from which the origin of the use certificate can be identified, which may be the hash of K_tcnant__Sj_gnpu|) is presented to the second security context 7 by the application server. The use certificate is presented each time the application server wishes to use K_(enant The application server may send a request message, specifying the use of the tenant key that is required, together with the use certificate. For example, the application server may send a message to the second security context 7 comprising a data file, a request that the second security context 7 encrypt the data file with the tenant key, a use certificate, the and the information from which the origin of the use certificate can be identified, which may be the hash of K_tenant__slgnpub.

The second security context 7 may have access to a number of public keys, for example associated with different tenants. The keys may be stored at the second security context 7, or outside the second security context 7 in an authenticated manner. The second security context 7 identifies the correct signature verification key ^tenant-sign_pUb> based on the information from which the origin of the use certificate can be identified, which may be the hash of K_tenant__sig_npub. The signed use certificate is then verified at the second security context 7. The second processor 19 in the second security context 7 is configured to verify the signature. The signature is verified using the public half of the signing key Kt_enant.si₈n pub· The second processor 19 is configured to

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 perform a signature verification algorithm that, given the signed use certificate and the public key Kt_enant._Sign pub either accepts or rejects the message's claim to authenticity. If the signature is verified successfully the method progresses to step S1105. If not, an error is returned to the first security context 5, for example a message is sent stating “AccessDenied”. At this point, the communication between the first security context 5 and the second security context 7 is terminated.

In step S1105, the second security context 7 validates the use certificate with respect to the access control list stored in the second device memory 11. As part of this step, it is validated that the use certificate has not expired. It is also validated that the use certificate allows the operation that is requested, in other words that the use certificate matches the credential associated with the permission corresponding to the operation, or use, which is requested. The second processor 19 is configured to validate the use certificate. If the certificate is invalid, for example because of tamper indicated by the certificate signature, expiry or if the ACL does not allow the operation requested, then the second security context 7 may return an error message to the application server for example.

The second security context 7 identifies the securely stored tenant key corresponding to the use certificate from the reference to Ktenant in the use certificate.

The second security context loads the securely stored tenant key and ACL. Loading the tenant key involves making it available to the processor in the second security context 7, i.e. sending the tenant key and ACL to the device memory 9, if not already present. This may involve loading the tenant key and ACL from a non-volatile memory device outside of the second security context 7, decrypting the key and making it available to the second processor for example.

At this point the second security context 7 validates the use certificate and that the expiry time has not elapsed with respect to the reference time source. The second security context 7 can use the use certificate to enable the matching permission by validating that all fields given are as expected.

The second processor 19 is configured to confirm that the reference to Kt_enant and the reference to the second security context 7 match those in ACL corresponding to the

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 key Ktenant and stored in the second device memory 11. In an embodiment, the reference to Kt_enant is the hash of K_tenant. In an embodiment, the reference to the second security context 7 is the hash of the public half of the identity key K_{aD pub}

If the reference to K_teTOm and the reference to the second security context 7 match those in ACL, it is confirmed whether the validity period has expired. If not, an error is returned to the application server, for example a message is sent stating “CertificateExpired”.

The second processor 19 is further configured to confirm that the validity period has not expired with respect to the referenced time source 2. In an embodiment, the second security context 7 requests the current time stamp from the time source 2 corresponding to the reference in the use certificate. The second transceiver 15 in the second security context 7 is configured to send a request message to the time source 2. In response to the request message, the time source 2 generates a message including the current time stamp. The message is signed with the private half of the identity key of the time source, Ktsid _Ρην The third processor 41 in the time source is configured to cryptographically sign the message with the private half of the identity key of the time source, K_Tsid priv- The signed message is then sent to the second security context 7. The transceiver 39 in the time source 2 is configured to send the signed message to the second security context 7. The message may also comprise the current configuration information of the time source.

The second security context 7 then verifies the response signature using the public half of the identity key of the time source, K_Tso pub The second processor 19 in the second security context 7 is configured to verify the signature. The signature is verified using the public half of the identity key of the time source, K_Tso pub· The second processor 19 is configured to perform a signature verification algorithm that, given the signed message and the public key Ktsid pub either accepts or rejects the message s claim to authenticity.

If the signature is verified successfully the current time stamp is compared to the expiry time contained in the use certificate to determine whether the use certificate has expired. If not, an error is returned to the application server, for example a message is

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 sent stating “AccessDenied. At this point, the communication between the first security context 5 and the second security context 7 is terminated.

If the airrent time stamp received from the time source 2 is prior to the expiry time contained in the use certificate, then the method progresses to step S1106. If not, an error is returned to the application server, for example a message is sent stating “CertificateExpired”. The application server may then request the first security context 5 to provide a new certificate or simply terminate the operation.

In step S1106, the second security context 7 grants the use of the cryptographic key, Ktenant, on the condition that the use certificate is valid and not expired. If the certificate is valid and the expiry time has not yet elapsed, the cryptographic operation is allowed to proceed. The second security context 7 can now use the tenant key within the validity period. The second security context 7 activates the necessary permissions within the ACL corresponding to the requested use and checks the ACL for an active permission that allows the request. The second security context 7 then performs the requested operation and returns the result to the application server. For example, in the case that the application server has sent a message comprising a data file and a request that the second security context 7 encrypt the data file with the tenant key, the second security context 7 encrypts the data file with the tenant key and returns the encrypted data file to the application server.

At or before the point of expiry, the tenant can generate a new certificate to extend the expiry time of the key. Failure to regenerate a new certificate results in the key becoming unusable by the service provider. In an embodiment, the second security context 7 notifies the first security context 5 before or at the point of expiry of the use certificate. A new use certificate can then be generated at the first security context with a later expiry time. Alternatively, the second security context 7 notifies the application server before or at the point of expiry of the use certificate. The application server then requests a new use certificate from the first security context 5. Alternatively, the use certificate is read by the application server, which requests a new use certificate from the first security context 5 before or at the point of expiry. Alternatively, a further component, which may be owned by the tenant or the service provider may monitor certificates and raise a request for certificates that are about to or have expired.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

Thus the tenant system 1 effectively leases the cryptographic key K^ant to the service provider system 3 in a secure manner. The use of the key by the service provider is cryptographically restricted by an expiry time in reference to a trusted time source, as set by the owning tenant in the use certificate. The ACL specifies that the use of the cryptographic key Kt_enant is only granted when an unexpired use certificate is provided. Furthermore, the tenant can restrict how the key is used by the service provider.

Figure 13 is a schematic illustration of a method of controlling use of a cryptographic key, Ktenant according to an embodiment of the present invention.

This method shown corresponds to the case where the tenant key is being used at the second security context 7, and the expiry period elapses. The second security context 7 or the application server may periodically compare the current time obtained from the reference time source 2 to the expiry time in the use certificate, and from this information determine whether the certificate has expired. Alternatively, the second security context 7 may determine that the certificate has expired only when a request using an expired certificate is made. The second security context 7 may notify the first security context 5, the application server or a further device that the key has timed out.

The first security context 5 is configured to send a request message to the time source 2 in response to the notification, requesting the current time stamp as described in relation to step S1202 above.

The time source 2 is configured to generate and send a message comprising the time information as described in relation to steps 1203 to 1205 above.

The message is verified at the first security context 5 as described in relation to step S1206 above and a use certificate is generated as described in steps S1207 to S1208 and signed as described in step S1102 above. The use certificate is sent to the service provider as described in step S1103 above, and forwarded to the second security context 7 and verified at the second security context 7 as described in relation to step S1104. The stored tenant key corresponding to the use certificate is loaded and the current time stamp obtained from the time source 2 and validated as described in relation to S1105 above. Loading the tenant key involves making it available to the processor, i.e. sent to the device memory 9 if not already present. This may involve

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 loading the tenant key from non-volatile memory device outside of the second security context 7, decrypting the key and making it available to the processor. The key is then activated as in step S1106 above.

Figure 14 is a schematic illustration of a method of controlling use of a cryptographic key, Kienam according to an embodiment of the present invention.

In the initial step, the first security context 5 obtains the generation certificate of the time source, status and time from a time source 2. The generation certificate and time stamp signature are then verified at the first security context 5. This is described in relation to Figure 10 and steps S1201 to S1206 above.

If the generation certificate and time stamp signature are validated, then the method progresses on to the next step, “generate authorisation certificate”. If not, an error is returned.

The use certificate is generated as described in steps S1207 and S1208 above. The use certificate is then sent to the second security context 7, as described in relation to step S1103 above.

The steps outlined above define a system that is secure against attack by third parties as well as a malicious service provider, as all key material is encrypted between the tenant and the second security context 7, the tenant key material is stored in a secure, tamper-resistant format, the tenant keys can only be used when authorized by a tenant in a way that is cryptographically enforced, and component spoofing is prevented by the presence of an identity key K_(D and generation certificate signed by a manufacturer that is trusted by both the tenant and the service provider.

The above described key transfer method and method of controlling the use of the key mean that the tenant can securely lease cryptographic keys to a service provider who is hosting a shared cryptographic infrastructure of known origin. The use of the tenant key remains under the control of the tenant in a cryptographically protected manner.

As part of the key transfer method, the tenant generates a key inside a self-hosted, secure appliance, the first security context 5.

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264

The tenant's secure appliance, the first security context 5, validates that the cryptographic infrastructure hosted by a chosen service provider, the second security context 7, adheres to a set of rules acceptable to the tenant by validating that the infrastructure is built using a trusted manufacturer, configured in a way acceptable to the tenant and suitable for leasing a key to. Step S205 described in relation to Figure 2(a) above allows validation that the infrastructure is built using a trusted manufacture. If the signature of the generation certificate C^d is verified, then it is validated that the second security context is built by a trusted manufacturer. Step S215 described in relation to Figure 2(b) above allows validation that the infrastructure is configured in a way that is acceptable to the tenant and suitable for leasing a key to.

In addition the tenant and service provider’s cryptographic infrastructure use a trusted time source 2, which may be hosted by a third-party, to enforce key expiry times set by the tenant.

The above described devices and methods enable a multi-tenanted, hosted cryptographic solution to be provided by a CSP. The system is designed to be secure against attacks by both third-parties and the CSP itself. A tenant remains in control of their key, providing usage authorizations, which are cryptographically enforced, for use inside a CSP-hosted system. It allows both multi-tenancy of HSMs and assurance for keys transferred into a service provider's cryptographic infrastructure. More than one tenant can use a single cryptographic appliance, however the security of a tenant’s key is protected from both third parties and the service provider.

The above described devices and methods allow a tenant to verifiably lease cryptographic keys to be used inside a service provider's cryptographic infrastructure in a way that is inaccessible to the service provider and third parties, whilst maintaining control over the use of the keys once inside a CSP security context. This enables the secure leasing of cryptographic keys between a tenant and a service provider, with the ability to restrict the use of a tenant key for use inside a service provider HSM for a set time period, and providing resistance to malicious or rogue administrator attacks. It allows leasing of keys to a CSP in a cryptographically enforced manner. It allows service providers to host multi-tenanted services that are robust to third-party attack whilst also being robust against a malicious CSP. Cryptographically enforcing that the

Date Reçue/Date Received 2021-06-25

WO 2017/134445 PCT/GB2017/050264 use of a tenant key is controlled by a tenant ensures that the use of keys is now completely controlled by tenants, making the service secure against attacks and enabling the sharing of infrastructure with others.

The above described devices and methods allow a CSP to provide their tenants use of their cryptographic infrastructure for applications such as payments, security and regulation.

The second security context 7 is part of a service provider system, enabling the service provider, for example a CSP, to host cryptographic services via hardware security modules, HSMs, on behalf of their customers.

While certain embodiments have been described, these embodiments have been presented by way of example only, and are not intended to limit the scope of the inventions. Indeed the novel methods and apparatus described herein may be embodied in a variety of other forms; furthermore, various omissions, substitutions and changes in the form of methods and apparatus described herein may be made without departing from the spirit of the inventions. The accompanying claims and their equivalents are intended to cover such forms of modifications as would fall within the scope and spirit of the inventions.