Communication method with encryption key escrow and recovery

i

COMMUNICATION PROCESS WITH SEQUESTRE

AND RECOVERY OF THE ENCRYPTION KEY

DESCRIPTION

Technical area

The present invention relates to a communication method in which encryption key sequestration and recovery operations are provided. These operations make it possible to guarantee to one or more previously determined organization (s) (for example a security administrator of a company network, a trusted third party, even, in certain cases, the users of a system). codes themselves), the possibility of recovering, if necessary, the session key used in the communication, and this from the data exchanged. The possibility of recovering a session key may arise from a need for legal interception or key recovery within a company.

The invention finds application in secure communications.

State of the prior art

Essentially, two families of key escrow / recovery techniques are known, guaranteeing one or more escrow authorities the ability to reconstitute, from the data exchanged during a communication between two interlocutors or entities a and b, the session key used, in order to decipher this communication. These two families of techniques have the advantage of being able to be implemented without any data exchange having to be carried out at each communication between the entities and the sequestration authority (s) (so-called off process). line).

Family 1: archiving of static key distribution keys with a escrow authority

This family of techniques applies to systems in which the establishment of a session key between the interlocutors uses a key exchange protocol based on the possession, by one of the interlocutors (for example b) of a static secret key (that is to say not renewed at each session). The secret key used by b in the key exchange protocol is archived with a escrow authority (or distributed among several escrow authorities). The possession of this secret allows the sequestration authority (or authorities) to reconstitute, if necessary, any session key exchanged between a and b from the messages of the protocol for establishing this key. An example of this key escrow and recovery method is provided in the article A Proposed Architecture for Trusted Third Party Services, by N. Jefferies, C. Mitchell and M. Walker, published in Lectures Notes in Computer Science 1029, Cryptography Policy and Algorithms Conference, pp. 98-104, Springer Verlag, 1996. It is one of the main methods, belonging to this first family of techniques, which has, until today, been considered in Europe.

Family 2: recovery of dynamic encryption keys (session keys) using legal fields

This second family of techniques does not call, unlike the previous one, the prior archiving of static secret keys used for the exchange of session keys, but the insertion, in the messages exchanged between a and b, when a secure communication, one or more legal fields containing, in a form intelligible only to a receiver authority, information on the session key KS. The key KS (or information on this key) can for example be encrypted under the public key RSA of a receiver authority. The Secure Key Recovery (SKR) protocol, proposed by IBM, is part of this family of techniques.

These two families of techniques have certain drawbacks for securing open applications that it is desired to be able to use between interlocutors located in different countries or different jurisdictions, such as for example secure messaging. When a secure application is likely to be used for international communications, it is desirable that two conditions be met:

(i) each country must be free to set up or not, for communications which concern it, a key escrow / recovery system for this application;

(ii) in each country where a key escrow / recovery system is in place, the authorities empowered to recover, if necessary, the session keys used to encrypt an international communication must be able to do so without having to cooperate, for each interception. , with the authorities of other countries.

However, the known techniques described above do not meet or do not meet these conditions:

- for the methods of the first family, when the method of distribution of the session key considered relates to public key encryption (in particular the RSA encryption used for this use in a very large number of security products), the recovery of the key session of a communication is only possible, without international cooperation, in the country where the secret key used for key distribution has been archived. This difficulty has led some authors (cf. the article by N. Jefferies et al. previously cited) to recommend key escrow / recovery systems based on a more symmetrical key exchange method, related to the Diffie-Hellman scheme. These systems satisfy the previous condition (ii) and could perhaps, subject to certain adaptations constrained conditions, satisfy (i), but they introduce strong points on the key distribution method used, which lead in particular to excluding the use of the RSA algorithm.

- For the processes of the second family, key recovery in the recipient country, from legal fields, is based on the implementation in the issuing country of a key escrow / recovery technique adapted to the recipient country, namely the issuance of intelligible legal fields for the escrow authorities of the recipient country. This constraint is in contradiction with the previous condition (i).

DE Denning's article Descriptions of Key Escrow Systems published in Communications of the ACM, vol. 39, No. 3, March 1996, as well as the article by DE Denning and DK Branstad entitled A taxonomy of Key Recovery Encryption Systems, published in Communications of the ACM, vol. 39, n ° 3, March 1996,

<td>give</td><td>a description</td><td>and</td><td>analysis</td><td>comparative</td><td>of</td>

<td>more than</td><td>thirty systems</td><td>of</td><td colspan="3">escrow and recovery</td>

<td>key.</td><td></td><td></td><td></td><td></td><td></td>

<td>We</td><td>can stick</td><td>at</td><td>two examples</td><td>illustrated</td><td>sure</td>

<td colspan="3">figures 1 and 2 attached</td><td></td><td></td><td></td>

In Figure 1, first of all, we see two entities a, b each equipped with cryptology means not shown and each provided with an identity Id_at, Id_b, a public key and a secret encryption key, respectively P_at, P_b, and S_at, S_b, as well as a C certificate_at, VS_b ; we see, moreover, two sequestration authorities T_at and T_b associated with the two entities a and b, these two authorities each archiving the secret keys S_at, S_b associated entities as well as their C certificates_at or C_b. The certificates attest to the correspondence between the secret key and the public key and that the secret key has indeed been archived. The certification authority is not shown in this figure.

The certificate may conform to ITU-T Recommendation X509.

The method of communication between these different means comprises the following operations:

A) entity a, which is supposed to engage in a session of transmission of a message M:

• checks the validity of C certificates_at and C_b, • produces a session key KS implementing a pseudo-random generator not shown, • uses its cryptology means to encrypt the session key KS with the public key P_b of the other entity and encrypt the message M with the session key according to a symmetric encryption algorithm, • transmits its identity ID_at or its certificate Ca, 'the encrypted session key P_b(KS) and the encrypted message E_KS (M),

B) entity b, to which the transmission is intended:

• checks the validity of C certificates_at and C_b, • retrieves the session key KS using its secret key S_b, • decrypts the message M using the session key KS.

In such a process, the receiver T_bcan, if desired, also recover the session key KS using the secret key S_b that it has archived and can thus recover the transmitted message.

This method has a drawback. Indeed, if the receiver T_b can retrieve the KS session key (because it archived the secret key T_b) and can thus recover the transmitted message, this is not the case for the receiver T_at, because it does not have the secret key S_b. It is then necessary to suppose a cooperation between the authorities of sequestration T_at and T_b, which is not obvious in the case of international communications.

This difficulty arises, in particular, from the fact that the key exchange process calls for asymmetric encryption-decryption using a pair of respectively public-secret keys, such as for example in RSA encryption. Some authors have advocated more symmetrical processes, akin to a so-called Diffie-Hellman protocol. This process is illustrated in Figure 2. There are substantially the same means as in Figure 1, namely the two entities a and b and the two sequestration authorities T_at and T_b. The parameters of the Diffie-Hellman protocol consist of a large prime number p, called modulus, and a generator number g. The two escrow authorities T_at and T_b agreed on these numbers p and g. The secret key S_at of a is a secret exponent a which is archived in T_at and the public key of a is P_at= g^at. Certificate C_at contains the public key P_at= g “. The same goes for entity b: (S_b= P, P_b= g^(J) .

To send a message to entity b, entity a generates a session key KS and addresses to b:

• his certificate C_at (which contains P_at= g “), • the session key encrypted by an algorithm E using the key g“^p, (Eg^aP(KS)), • the message encrypted by the KS session key (E_KS(M)).

Knowledge by T_at of a and the public key P_b= g^p of b allows T_at to calculate (g^p)^at= g^Pa. The same for T_b who can calculate (g^at)^p= g^ap. Thus, g^aP is it shared by a and b.

Each authority T_at or T_b can therefore find the session key (KS) and therefore recover the message (M).

But, here again, this diagram assumes an agreement between the parties.

The object of the present invention is to remedy these drawbacks by proposing a method which does not require any agreement between the communicating parties, the recovery of the session key and of the message being carried out solely on the basis of the data exchanged in the communication.

Presentation of 1¹ invention

Specifically, the subject of the invention is an encrypted communication method with sequestration and recovery of the encryption key, implementing:

- a first entity (a) comprising first cryptology means (MC_at) and provided with a first identity (Id_at), a first public key distribution key (P_at) and a first secret key distribution key (S_at) corresponding to said first public key (P_at) ,

- a second entity (b) comprising second encryption means (MC_b) and provided with a second identity (Id_b), a second public key distribution key (P_b) and a second secret key distribution key (S_b) corresponding to said second public key (P_b), this process comprising:

i) a preliminary phase of establishing a session key (KS) phase in which at least one of the entities (a, b) produces a session key (KS) and forms a cryptogram consisting of this key encrypted by the public key (P_b, P_at) of the other entity, the other entity (b, a) decrypting said cryptogram using its secret key (S_b, S_at) and covering the session key (KS), ii) a message exchange phase (M) in which the entities (a, b) form cryptograms E_Ks (M) consisting of messages (M) encrypted by the session key (KS) established in the preliminary phase, each entity decrypting the cryptogram that it receives using the session key (KS) and thus covering the message which was sent to it, this method being characterized by the fact that • it also implements at least one sequestration authority (T_at, T_b) associated with one of the entities (a, b), this authority archiving the secret key (S_at, S_b) of the associated entity (a, b), • in the preliminary phase, the entity (a, b) which produces the session key (KS) implements a pseudo-random generator (PRG_at, PRG_b) known to the receiver (T_at, T_b) associated and initializes this pseudo-random generator using its secret key (S_at, S_b) and an initial value (VI) deduced, by an algorithm known to the escrow authority (T_at, T_b), appropriate data.

ίο

According to one embodiment, the receiver authority (T_at, T_b) associated with the entity (a, b) which produced, in the preliminary phase, the session key (KS), implements a pseudo-random generator identical to that of the associated entity (PRG_at, PRG_b), initializes this generator with said initial value (VI) and the secret key (S_at, S_b) of the associated entity (a, b) that it has archived, and thus recovers the session key (KS).

According to another embodiment, the escrow authority (T_b, T_at) associated with the entity (b, a) which did not produce, in the preliminary phase, the session key (KS), decrypts the cryptogram of the session key (P_b(KS), P_at(KS)) using the secret key (S_b, S_at) of the associated entity (b, a) which it has archived and thus recovers the session key (KS).

As for the initial value VI, it can either be deduced from the data exchanged between the entities a and b in the preliminary phase of establishing the session key or be obtained by successive tests from data likely to take a given number of values. , this number being sufficiently small for the time taken by the escrow authority to be compatible with the envisaged application.

As indicated in the introduction, the escrow authority can be an authorized third party, or a security administrator of a corporate network, or the user himself (the escrow is then a self-receiver).

Brief description of the drawings

<td>- figure 1, already</td><td>described, illustrates</td><td>a</td><td>process</td>

<td>known asymmetric</td><td>r</td><td></td><td></td>

<td>- figure 2, already common symmetric</td><td>described, illustrates r</td><td>a</td><td>process</td>

<td colspan="2">- Figure 3 illustrates schematically according to the invention.</td><td>a</td><td>process</td>

Description of particular modes of implementation

The method of the invention can be described by first specifying certain initial conditions, then specifying the procedures developed in the user's encryption means and finally describing the procedure for recovering the key.

A. Initial conditions

The secret key S_at of the public key key encryption system used by entity a for session key establishment purposes is archived with the escrow authority T_at. Delivery to a, by a certification authority AC designated in advance by T_at, a C certificate_at attesting to the correspondence between the identity Id_at of a and the public key P_at (for example a certificate conforming to recommendation X509 of the ÜIT-T), must be subject to this archiving. The possession by a of a certificate from AC proves that the archiving with T_at of the secret key S_atcorresponding to the public key P_at has actually been achieved. In practice, the certification authority AC and the receiver T_at can be the same organization, or two separate organizations having entered into an agreement. Generation of the secret key S_at can, depending on the case, be carried out by user a or by third party T_at.

B. Procedures in the user's crypto facility

By means of cryptology of noted MC is meant_at, the software and hardware resources implementing the cryptographic calculations for establishing a session key and for encrypting a during a secure communication. For example, the client software of a secure messaging system can be considered as a means of cryptology.

For the MC cryptology medium_at user has either in accordance with the third party escrow service offered by T_at, it must meet the following conditions:

i) Performing MC encryption functions_at(establishment of a session key, encryption) must be subject to the presence of a C certificate_at from an AC certification authority designated by T_at and the secret key S_at corresponding. MC encryption_at must verify not only that the C certificate_at is valid, but still that there is indeed a correspondence between the secret key S_at and the public key P_at contained in T_at. These checks are necessary to ensure that the third party of receiver T_at is able to recover session keys processed by MC_at.

ii) The key generation process implemented in MC_at -typically the key generation algorithm used to generate a KS session key when a initiates a secure session with a party b- must be a known GPA pseudo-random generator of T_at, and whose seeds, i.e. the inputs from which the values ​​produced by the generator are calculated, are made up:

- the secret key S_at (or, according to a variant, of a function H (S_at) of this key);

- from an initial value VI deduced, by an algorithm known to T_at, variable data contained in the unencrypted part of the communications between a and its interlocutors (for example the date and time), or a counter managed inside MC_at.

The pseudo-random generator must meet the following conditions:

i) it should be easy to deduce the output value of this generator (typically the session key KS) from S_at (or H (S_at)) and the initial value VI. According to a preferred embodiment of the invention, the size of the initial value VI can be limited between 20 and 40 bits of effective size, so that, when the secret key S_at is known, the recovery of the output value of the generator remains possible by exhaustive search even when the exact value of VI is lost, ii) it must be difficult to predict information on S_at (or on H (S_at)) from a set of VI values ​​and GPA (S_at, VI) or GPA (H (S_at), VI) corresponding, iii) it must be difficult to predict information relating to the GPA outputs (S_at, VI) or GPA (H (S_at), VI) for the different values ​​of VI when the value of S_at (or H (S_at)) is unknown.

C. Key recovery procedures

There are two separate procedures for retrieving the KS session key used to encrypt a secure communication between user a and party b, through T_at or an authority authorized to access the secret S_at archived by T_at :

i) If the session key KS is produced by b and received by a encrypted using the public key P_at of a, then T_at can recover the key KS by decrypting using the archived secret S_at the cryptogram P_at(KS) transmitted in the key distribution protocol.

ii) If the KS session key is produced in the MC encryption means_at from a and sent to b encrypted under the public key P_b of b, then T_atcan recover in the data exchanged in clear between a and b the initialization value VI and reconstitute the value of KS using VI and the archived value of S_at, by performing the calculation KS = GPA (S_at, VI) or KS = GPA (H (S_at), VI). In the event that VI and the contents of a counter, or in the event that the effective size of VI is limited and for some reason VI cannot be retrieved from the clear data, it remains possible for T_at to recover the session key KS by carrying out an exhaustive test of the possible values ​​of VI, and by checking for each one if the value KS = GPA (S_at, VI) or KS = GPA (H (S_at), VI) obtained is the correct one.

By combining the elementary procedures i) and ii) defined above, T_at remains capable of recovering the session key in the event that a more complex session key establishment protocol is used between a and b. We can consider, by way of example, the following protocol: b generates a secret value KSI and transmits it to a encrypted under the public key P_at of a; a generates a secret value KS2 and transmits it to b encrypted under the public key P_b of b; a and b calculate the session key KS, equal to the exclusive OR of the values ​​KSI and KS2 (KS = K1 XOR K2). For a protocol of this kind, T_at would be able to recover KSI using procedure i) defined above and recover KS2 using procedure ii), and therefore, from these two values, recover KS.

The method which has just been described can be implemented according to variants in which the information that constitutes the secret key S_at is not archived with a single entity T_at, but divided into shares archived with separate third-party receivers.

For example, the secret key S_at of a can be made up of a secret RSA exponent d. This secret can be split into two parts dl and d2 such that dl + d2 = d. Two Receivership Authorities T_at and T '_at, respectively responsible for archiving dl and d2 (and the public module n_at of a), are capable of:

• to check, without revealing their part of the secret d, that they are indeed able to calculate the secret function associated with the key S_at ; To do this, it suffices for each of them to calculate, modulo n, the power of the input value determined by its share, 5 then for the values ​​obtained to be multiplied between them modulo n_at, • to recover a session key KS from the data of the protocol for establishing this key (by revealing, if necessary, to the other third party or to an interception authority their share of the key S_at) .