Multicast transmission method by decrementing an associated counter

The invention relates to the field of so-called "multiple access" communications networks, and more specifically the transmission of data between nodes of such networks.

Here, the term "multiple access network" is understood to mean a network in which at least one structure having an access bus function (in terms of resources and / or connections) is defined and to which a selected group of his knots.

Moreover, here "data transmission" is understood to mean any mode of transmission, whether it is a "point-to-point" mode transmission (or "unicast" transmission), or a transmission in "point-to-multipoint" mode (of "multicast" type if the data is intended only for a subgroup of a group of nodes, or of "broadcast" type if the data is intended for all nodes of a group), or a so-called "multipoint-to-multipoint" transmission.

As is known to those skilled in the art, the transmission of data between an input node, belonging to a group coupled to one of the access buses of a multiple access communications network, and one or more other destination nodes, belonging to the same group, requires association with the data, by stacking, either of connection identifier, for example in the form of MPLS (for "Multi Protocol Label Switching") labels, to differentiate, first of all the multiple access resources (or access bus) of the network (by an "external" or primary label), then a connection between nodes among all the connections supported by the access bus to which these nodes are coupled (by an "internal" or secondary label, or an address field according to the technology of the client network), or MAC addresses of the nodes that define the transmission path between the concerned nodes.

In the first case, when it is desired to provide connectivity between the N nodes of a group of the network, using the multiple access resources of the access bus to which they are coupled, it is necessary to define Nx ( N-1) / 2 sub-connections multiplexed together within the multi-access connection. This therefore requires processing in each node of the primary label to route the bus, then the secondary label of each bus data to process each connection. This limits the applications of the point-to-multipoint and multipoint-to-multipoint transmission modes because even though the node structure (ie the switching plan) is simplified by bus switching, control remains the same as that used in a packet-switched network.

In the second case, a technique called "destination deletion" (or "destination stripping") must be implemented. This technique consists in extracting the data received at the destination node which is designated by the MAC recipient address associated with them, in particular in Ethernet and RPR networks (for "Resilient Packet Ring"). The main drawback of this technique lies in the fact that it takes place in an absolute and flat address space (that is to say in which the MAC addresses are hard-coded, universally and not relatively, and without hierarchy). With this type of technique, once all MAC addresses have been assigned to nodes in a network, no new nodes can be added to the network. In addition, multicast or broadcast packets must have a specific address to differentiate them from unicast packets, which further limits addressing capabilities.

The invention therefore aims to improve the situation.

It proposes for this purpose a method of data transmission between nodes of a group of nodes coupled to one of the access buses of a multiple access communications network.

This process is characterized by the fact that it consists of:<ul id="ul0001" list-style="dash" compact="compact"><li>whenever an input node of the group receives data intended for at least one of the other nodes (recipients) of the group, determining the number M of nodes in this group through which the data must pass to reach its destination, including the recipient node (s), and which together define a transmission path from said input node, then to assign to a dedicated field, which is associated with the data, a value representative of a number, this number being equal to M before transmitting said data, via the access bus, at the first node of the path,</li><li>then at each node of the path, to decrement by a unit the number that is associated with the received data in order to obtain a new (decremented) value of the dedicated field, and<ul id="ul0002" list-style="none" compact="compact"><li><img file="EP1575211A2_D0001.tif" /> in the presence of a new value (decremented) equal to a chosen comparison value, to extract the data from the path, after a possible copy, or</li><li><img file="EP1575211A2_D0002.tif" /> in the presence of a new value (decremented) different from the comparison value chosen, to be replaced by this new value the previous value of the dedicated field associated with the received data, then to transmit to the next node of the path the data and the new value ( decremented) of the associated dedicated field.</li></ul></li></ul>

Preferably, the input node is used to add the field dedicated to the received data. This field thus makes it possible to avoid having two levels of addressing in each node: the value of the field is thus used to replace the secondary addressing level (secondary label or client address) and only the addressing of the bus (label primary) remains. This saves processing at the control plane level to maximize the benefits of point-to-multipoint and multipoint-to-multipoint transmission modes.

The value chosen, to which is compared the new value resulting from a decrementation, is for example the null value, by default. But, this chosen value can be also different from zero.

Moreover, it is possible to proceed in the input node to add to the received data (arranged in the form of packets or frames) an additional field that can take a first value signaling a transmission in point-to-point mode. a second value signaling a broadcast-type point-to-multipoint transmission and / or a third value signaling a multicast point-to-multipoint transmission. As a variant, it is possible to reserve first specific values ​​for the transmission of the received data in point-to-point mode (or unicast), and second specific values ​​for the transmission of the data received in broadcast mode. In either case, when the data must be transmitted in broadcast mode and as long as the value of the dedicated field is different from the chosen comparison value, it is possible to proceed at each node of the path to the duplication (or copy) data and associated fields for retrieval and transmission after replacement of the value of the associated dedicated field by the new value (resulting from the decrementation at the node).

When the transmission of the received data (arranged in the form of packets or frames) must be done in multicast mode, which corresponds to a transmission to destination nodes belonging to a subgroup (or domain) of the group, it is possible to proceed each destination node has a new determination of the number M of nodes of the next transmission path that the data must follow to arrive at the next destination node. Then, the dedicated field that is associated with the received data is assigned a value equal to M, before transmitting them to the first node of the following path.

In a first variant, it is possible to determine for each destination node Ni of the subgroup, the number Mi of nodes in this group that define the transmission path that the data must follow to reach it, then we associate to the data to be transmitted a dedicated field for each destination node Ni and we assign a value representative of the number Mi to each of the dedicated fields, then we transmit the associated data and associated fields, via the access bus, at a first node of the path, then, at each node of the path, the value of each dedicated field associated with the data received is decremented by one unit in order to obtain new values, and in the presence of a new value equal to the chosen comparison value (for example zero), duplicate (or copy) the data and associated fields for extraction and transmission after replacing the values ​​of the dedicated fields associated with them by the new values.

A variant of this first variant can consist in that the value (Mi) of each field, associated with a destination node Ni, represents the number of jumps after the node Ni-1. In this case, only the first field (Mi) is decremented in a node Ni, and when its value becomes zero, the frame is duplicated (or copied) in order to extract it and transmit its copy, but without this first field ( Mi), at the next node Ni + 1 of the transmission path identified by the following field (Mi + 1) now placed at the top of the stack.

In a second variant, the comparison value chosen for extracting a packet locally in a destination node Ni may be configurable. More precisely, it can be envisaged that this comparison value chosen varies from one recipient node to another within a given subgroup. Indeed, by default the value chosen is preferably zero, but in the multicast case, it is possible to define a particular comparison value per node Ni belonging to the subgroup, the initial value M remaining equal to the total number of hops until last node of the subgroup.

When the network is circuit-oriented transport layer (such as for example G.709 type circuits), the data constitute packets that are grouped within frames. In that case, we associate a field dedicated to each packet of a frame, the determined value of each dedicated field being a function of the destination node at which the packet is to be extracted, and at each node of the transmission path, the value of each dedicated field contained in the received frame is decremented by one unit in order to obtain new values, and in the presence of a dedicated field having a new zero value, we extract the packet that is associated with this dedicated field, while in the presence of a field with a new non-zero value, we replace its value with the new value, and then transmitting to the next node of the path the frame comprising the dedicated fields which are associated with the remaining packets and which comprise the new values.

In this case, the dedicated fields, which are associated with the packets of a frame, can be grouped in an orderly manner into a control packet, according to the scheduling of the packets associated with them within the frame. For example, the control packet can be placed just after the header of the frame, at the beginning of the frame part which is dedicated to the payloads included in the packets. Alternatively, the control packet may be placed at the end of the frame portion dedicated to the payload included in the packets. In another variant, the dedicated fields, which are associated with the packets of a frame, may be placed in an orderly manner in the header of the frame, in accordance with the scheduling of the packets associated with them within this frame. In all these situations, dedicated fields of variable or fixed size can be used.

Other features and advantages of the invention will appear on examining the detailed description below, and the attached drawings, in which:<ul id="ul0003" list-style="dash" compact="compact"><li>FIG. 1 schematically illustrates an example of a communications network making it possible to implement a data transmission method according to the invention,</li><li>FIG. 2 schematically illustrates a first exemplary frame adapted to the implementation of the invention in an OSPF-type GMPLS-type network,</li><li>FIG. 3 schematically illustrates a second example of a frame adapted to the implementation of the invention in a GMPLS network of OSPF type routing type,</li><li>FIG. 4 schematically illustrates a third example of a frame adapted to the implementation of the invention in a GMPLS-type network with OSPF-type routing, and</li><li>FIG. 5 schematically illustrates an example of a frame adapted to the implementation of the invention in a G.709 type circuit-oriented transport layer network.</li></ul>

The attached drawings may not only serve to complete the invention, but also contribute to its definition, if any.

The object of the invention is to enable the transmission of data between nodes of a multiple access communications network.

Here, the term "multiple access network" is understood to mean a network in which access bus structures are defined to which respective groups of nodes, possibly variable in time, are coupled.

Referring first to Figure 1 to describe an embodiment of a communications network R1 to implement the invention.

In this exemplary embodiment, the network is considered to be of the Generalized Multi Protocol Label Switching (GMPLS) type and uses an OSPF (Open Shortest Path First) type routing protocol. It is recalled that the OSPF protocol is an algorithm for routing the family of Internet protocols (IP for "Internet Protocol"). But, of course, the invention is not limited to this type of network. It will also concern, as will be seen below, the circuit-oriented transport layer networks, and in particular the G.709-type networks, as well as all the networks using any bus-type resource by its client layer as a resource. multiple access.

The network R1 illustrated in FIG. 1 comprises, conventionally, a multiplicity of network equipment constituting nodes Ni (here i = 1 to 9, but it can take any value greater than or equal to two). These nodes Ni are connected to each other and responsible for routing between them data frames, so that they can be transmitted between at least one terminal (or server) of source communications CLj, belonging, for example, to a client j, and at least one destination communication terminal (or server) CLk, belonging, for example, to another client k.

The nodes Ni are for example peripheral routers (or "edge routers"), connected to one or more client terminals CLj and one or more other routers Ni ', or core routers (or "core routers") connected to other peripheral and / or core routers.

In the example shown, the node N1 is connected to two client terminals CL1 and CL2, the node N1 is connected to a customer terminal CL3, the node N8 is connected to a customer terminal CL4, the node N4 is connected to a client terminal CL5, and the node N5 is connected to another network R2, for example wireless network type (or WAN for "Wireless Access Network"), through an access server SA.

Furthermore, in a GMPLS type network, virtual circuits (or connections), called LSPs (for "Label Switched Paths"), are established between certain nodes Ni. These LSPs are based on a GMPLS control plane and require a (quasi) level 1 (in the so-called "TDM-LSP"), (quasi) level 2 (in the so-called "Layer 2 LSP "(Or L2-LSP)) or (almost) level 3 (in the case called" Packet LSP "). Within a GMPLS (R1) network, each LSP is designated by a label called GMPLS label (or G-label). As a result, LSPs are switched according to the values ​​taken by G-label labels.

In the network R1 according to the invention, at least a first level of LSP corresponding to the buses which cover it is defined. Each of these LSPs, which can be called bus-LSP, is identified by a G-label called primary label which serves to switch the bus in each node Ni. Said bus-LSPs are therefore in reality:<ul id="ul0004" list-style="dash" compact="compact"><li>either packet-LSPs, if the granularity of the resource is of level 3 (for example IP packet with a header of shim (or "shim header") MPLS),</li><li>either L2-LSPs, if the granularity of the resource is level 2 (for example an Ethernet frame with a G-Label label as represented in FIG. 2),</li><li>or else TDM-LSPs, if the granularity of the resource is level 1 (for example a SONET / SDH circuit frame),</li></ul> but with the particularity of being seen by a client layer as a multi-access bus.

In the following it is considered that the granularity of the resource is of level 2, so that one speaks of circuit L2-LSPs.

It is recalled here that a bus can be of the point-to-multipoint type, when there is only one transmitter on it, and then have broadcast type connections, when all the nodes N0 are recipient nodes to retrieve a packet, or multicast type, when some recipient nodes Ni must retrieve a packet, but not all. A bus can also be of multipoint-to-multipoint type when several transmitters and several receivers (or recipients) are on it.

Moreover, a connection on a bus can be of the point-to-multipoint type in order to allow the transmission of packets to several other nodes connected to the bus in broadcast or multicast mode, but it can not be of the multipoint type. -multipoints because there is only one issuer involved.

In other words, the multipoint-to-multipoint or point-to-multipoint aspect does not occur at a transmitter node (or point) but at the bus level.

The data frames, originating from a source client terminal CLj, access the network via a peripheral node Ni, and are multiplexed in one of the L2-LSPs circuits by means of interworking functions (or "Interworking functions") implemented. in a network / user interface called UNI (for "User Network Interface"). Each L2-LSP circuit constitutes a network access bus structure R1. When an L2-LSP circuit (or access bus) is defined by N nodes Ni, or in other words when N nodes Ni, constituting a group, are coupled to the same circuit L2-LSP, it is defined within of this L2-LSP Nx (N-1) / 2 circuit under connections to provide connectivity between the N nodes N of the group.

The invention proposes a method dedicated to the transmission of data between the nodes N1 of a group of nodes, coupled to one of the access buses (here of the L2-LSP type) of a multiple access communications network. (here the R1 network of GMPLS type).

This method therefore applies to situations in which an input node (source), for example N2, coupled to a circuit L2-LSP defined by the nodes N1 to N5, has received from the client terminal CL3 (arrow F1) to which it is connected, via its UNI, data to be transmitted to at least one other client terminal, for example CL5 (arrow F2), which is connected to a destination node also coupled to said L2-LSP circuit, for example N4. Therefore, it is considered that the network R1 is capable of determining the path (or (portion of) circuit) of inter-node transmission Ni that must borrow the data to reach the destination client terminal (here CL5). This transmission path here consists of a portion of the circuit L2-LSP (materialized by thick arrows), defined by the nodes N2, N3 and N4.

The method according to the invention consists, in the aforementioned situation, in firstly determining the number M of nodes Ni, coupled to the circuit L2-LSP concerned, which define the transmission path that the data must follow from the client terminal. CL3 and received by the input node N2, in order to arrive at the destination node N4. Here, the nodes defining the transmission path from the input node (or source) N2 are N3 (intermediate node) and N4 (destination node). Therefore, in this example M is equal to two (2).

It is important to note that the transmission path is here defined between the first intermediate node (here N3), to which is connected the input node (here N2), and the destination node (here N4). But, of course, other definitions of transmission path can be envisaged. It can indeed be considered that the path is defined between the input node and the destination node. In this case, the number M retained is equal to the total number of nodes defining the transmission path decremented by one unit.

Once the number M has been determined, a dedicated field C, which is associated with the data to be transmitted, is assigned a value representative of the number M, before transmitting this data and the dedicated field C, via the L2-LSP access bus. at the first node (intermediate) of the transmission path (here N3). Preferably, in the following, and with the exception of the exception indicated, the value assigned to the dedicated field C is equal to the number M.

This determination is preferably carried out at the level of the input node (here N2), thanks to a dedicated "interworking" function (or IWF) of the UNI network / user interface, which is furthermore in charge of associating with data to transmit the G-label field whose value designates the L2-LSP circuit they must borrow.

In a GMPLS network using an Ethernet transport format, the data to be transmitted is usually integrated into a T-frame having a six-byte DA field (destination terminal address), a six-byte SA field (source terminal address), a Four-byte G-label field, a two-byte type field, a data field between forty six and one thousand five hundred bytes, and a four-byte FCS field ("Check Sum"). In a GMPLS network (R1) according to the invention, the UNI is responsible for adding to the frame the dedicated field C which initially has the value M. The latter occupies for example a byte. Such a frame (T1) is illustrated by way of example in FIG. The order of the different fields may vary.

Once the T1 frame is constituted, it is inserted on the circuit L2-LSP concerned which routes it to the first intermediate node (here N3) of the transmission path, which is connected to the input node (here N2). The node N3 reads the value M of the dedicated field C (here equal to 2), then it decrements by one unit this value in order to obtain a new value M '(here equal to 1).

It then compares the new value M 'with the value zero (0). If M 'is equal to zero (0), it extracts the frame T1 from the transmission path, and therefore from the circuit L2-LSP. On the other hand, if M 'is different from zero (0), it replaces the value M of the dedicated field C by the new value M', then it transmits to the next node (here N4) of the transmission path, always via the circuit L2 -LSP, the T1 frame having the new value M 'of the dedicated field C.

The dedicated field C can therefore be likened to a counter which, when its decremented value M 'is equal to zero, designates the destination node of a frame T1. This makes it possible to avoid using an address field in each frame T1 to designate its destination node. When several clients are connected to the same node of the network R1 (such as N1), the output UNIs must sort the extracted packets having a counter M to 0 to send them to the good client CL1 or CL2. This is done conventionally by analyzing the client addresses that remained "hidden" in the network R1.

In the illustrated example, M 'being equal to 1, the frame T1 is thus transmitted to the node N4, with its new value M'.

When the next node (here N4) receives the frame T1, it performs the same operations as the previous node N3 (reading the value M of the dedicated field C, then decrementing a unit to obtain M ', then comparing M' to the value zero (0), and extraction or transmission). As long as the value M 'of the dedicated field C (after decrementing) is not equal to zero, the node having received the frame T1 transmits it to the next node of the circuit L2-LSP to which it is coupled. In the example illustrated, the decrementation of the value M (equal to 1) of the dedicated field C, performed by the node N4, provides a new value M 'equal to zero. Therefore, the node N4 extracts the T1 frame from the L2-LSP circuit, in order to transmit it to the client terminal CL3 designated by the address field SA. In other words, the node N4 is here the destination node of the frame T1, even though said frame T1 did not include an address designating it as such.

It is important to note that in the above example, the chosen comparison value used to compare the decremented value M 'at a node of the transmission path is zero (0). But, it can be envisaged that this chosen comparison value is different from zero. In this case, and in the presence of a unicast transmission, the initial value of M must be equal to the number of nodes of the transmission path, plus a chosen value, for example equal to the comparison value. It is in this situation that the value of the field C is representative of the number M and not only equal to M.

The processing of the T1 frames (reading, decrementing, comparison and decision), described above, is preferably performed by a processing module MT according to the invention, implanted in (or connected to) each node Ni of the network R1.

This processing module MT can be realized in the form of electronic circuits, software modules (or computer), or a combination of circuits and software. It could even be implemented in an all-optical way, as described in the article by J. E. McGeehan et al, "All-optical decrementing of packet-time-to-live (TTL) field and subsequent dropping of zero-TTL Packet", Journal of Lightwave technology, vol. 21, No. 11, Nov. 2003, but for the purpose of avoiding infinite loops in the network (like any IP TTL field), as indicated in this article, but allowing the "routing" of a frame within an LSP bus.

In the foregoing, a situation has been described in which the T1 frame is transmitted in a point-to-point mode (or "unicast"). But, the invention is not limited to this single mode of transmission. It can for example be applied in conjunction with point-to-point and point-to-multipoint broadcast and / or multicast transmission modes.

For this purpose, it is possible to add to the received frame, in the input node (using its UNI interface), an additional (or complementary) field CC that can take either a first specific value signaling a bit-mode transmission. -a-point, either a second specific value signaling a transmission in broadcast mode, or a third specific value signaling a transmission in multicast mode. A T2 frame of this type is illustrated by way of example in FIG.

When the additional field CC is used to signal a packet to be transmitted in unicast or broadcast mode, it may, for example, consist of a single bit taking the values ​​0 or 1. In this case it is the set C + CC which consists of a byte (C being then defined on seven bits instead of eight previously). When the additional field CC is used to signal a packet to be transmitted in unicast, broadcast or multicast mode, it may, for example, consist of two bits taking the values ​​00, 01 or 11.

As a variant, it is possible to reserve, within a frame T1, specific first values ​​for the transmission of data received in point-to-point mode, and second specific values ​​for the transmission of the data received in broadcast mode.

In one or the other embodiment, when a T1 or T2 frame is to be transmitted in broadcast mode, the duplication (or copy) of the frame is performed at each node of the transmission path as long as the value M 'of the dedicated field C is non-zero, in order to extract it and to transmit to the next node its copy, of course after replacement of the value M of the dedicated field C by the new value M'. In other words, in broadcast mode the processing module MT is arranged to force the extraction and copying of the frame T1 or T2 even though the value of the dedicated field (or counter) C is not equal to zero.

The invention can also be applied to the multicast point-to-multipoint transmission mode. In this transmission mode, the frame must be transmitted to several destination nodes belonging to a subgroup (or domain) of the group of nodes defining an L2-LSP circuit.

In order to allow this type of transmission, at least three embodiments can be envisaged.

A first embodiment consists in performing in each destination node a new determination of the number M of nodes of the next transmission path that the data must follow to arrive at the next destination node. This can be done because all the nodes that belong to the same domain (or subgroup) know each other.

When a node does not have a UNI user interface, for example because it constitutes a core router, it is preferentially its processing module MT that determines the number M of nodes N1.

Once a new transmission path and the value M of the associated dedicated field C have been determined, at the node constituting both an input node and a destination node, the duplication (or copy) of the received frame in order to extract it and transmit its copy to the first node of the following path, of course after replacement of the previous value M of the dedicated field C by the new value M.

It is important to note that this first embodiment corresponds more particularly to the case where the chosen comparison value is the null value (0).

A second embodiment consists first of all in determining for each destination node Ni of the subgroup (or domain) the number Mi of nodes of this group which define the transmission path that the frame must follow to reach it. Then, to the received frame, at the input node, a dedicated field Ci is added to each destination node Ni and a value equal to Mi is assigned to each of the dedicated fields Ci. Such a frame T3 is illustrated by way of example in FIG. 4.

Next, the frame T3, via the relevant access bus L2-LSP, is transmitted to the first node of the transmission path. Then, at each node Ni of the transmission path, the value Mi of each dedicated field Ci is decremented by one unit in order to obtain new values ​​M'i, and when a new value M'i is zero, it is duplicated. (or copy) the frame T3 in order to extract it and transmit its copy to the next node of the transmission path, of course after replacing the values ​​Mi of the dedicated fields Ci by the new values ​​M'i. Of course, each time one of the fields Mi reaches the value zero (0), this field can be optionally removed to "display" the next non-zero value Mi, which represents the number of hops on the bus before the next extraction .

Alternatively, at the time of creation of the frame T3, each field Mi no longer contains the number of jumps between the source node and the respective destination node in the node Ni, but the number of jumps after the node Ni-1. Thus, only the first counter Mi is decremented in a node, and when the first counter Mi reaches the value 0, the frame T3 is duplicated (or copied) in order to extract it and transmit its copy, but without this counter E at the next node of the transmission path identified by the new counter Mj now placed at the head of the stack.

It is important to note that this second embodiment corresponds more particularly to the case where the chosen comparison value is the null value (0).

A third embodiment can be envisaged in which the chosen comparison value is configurable, and more precisely that it varies from one recipient node to another within a given subgroup.

It is possible to define a particular value Mi per recipient node Ni belonging to a given subgroup. Each value Mi is determined according to the position of the destination node Ni in the transmission path. The initial value M is always equal to the total number of hops to the last destination node of the subgroup. As an illustrative example, it is considered that M = 10, that the subgroup comprises four destination nodes N1 to N4, and that the destination nodes N1, N2, N3 and N4 are respectively placed in third, fifth, seventh and tenths. position on the transmission path. In that case, the first node N1 extracts the data identified as multicast by the additional field CC and associated with the decremented value M1 = 7 (the selected comparison value at the level of N1 is then equal to 7), the second node N2 extracts the data identified as multicast by the additional field CC and associated with the decremented value M2 = 5 (the chosen comparison value at the level of N2 is then equal to 5), the third node N3 extracts the data identified as multicast by the additional field CC and associated with the decremented value M3 = 3 (the chosen comparison value at the level of N3 is then equal to 3), and the fourth and last node N4 of the subgroup extracts the data associated with the value M4 = 0 (the selected comparison value at the level of N4 is then equal to 0).

As indicated above, the additional field CC can designate a packet to be transmitted in the multicast, broadcast and unicast modes when it consists for example of two bits. But when it consists of only one bit it can only designate a packet to transmit in broadcast and unicast modes. It is then necessary to provide another additional field CC 'to specifically designate a packet to be transmitted in the multicast mode.

This variant is advantageous because it offers a processing that requires only the decrementation of a single field, without the need to determine in each destination node Ni a new value to reach the next destination node.

Referring now to Figure 5 to describe an implementation of the method according to the invention in a network R1 circuit-oriented transport layer, for example G.709 type.

In this case, the access buses are G.709 circuits for transporting within T4 frames (see Figure 5) several packets or portions of packets, possibly encapsulated according to possibly different chosen protocols. Furthermore, the nodes N1 comprise, on the one hand, a G.709 frame circuit switch making it possible to "route" the G.709 frames to the other nodes of the network, and on the other hand, insertion modules and extracting the packets respectively in and out of the frames.

According to the invention, a dedicated field Ci is associated with each packet Pi to be transported by the same frame T4. The initial value Mi of each dedicated field Ci of a frame T4 is determined at the level of the input node (here N2), by the dedicated interworking function (IWF) of its UNI, as a function of the destination node Ni at the level of which the corresponding Pi packet is to be extracted.

It is important to note that dedicated fields Ci can have variable or fixed sizes.

In the example illustrated in FIG. 5, the frame T4 comprises an OH header (or overhead) comprising information and control fields, as well as a PD portion dedicated to the payloads comprising four packets P1 to P4. The four dedicated fields C1 to C4 associated with the four packets P1 to P4 are here placed in the PD part. They are more precisely grouped together, one after the other in the same order as the packets P1 to P4, so as to constitute a PC control packet (or counter vector). This PC control packet is here placed at the beginning of the PD part, just after the OH header and just before the first packet P1. But, the PC control package could be placed at the end of the PD part, right after the last P4 packet. It could also be placed in the OH header.

The PC control packet is preferentially substituted for the G.709 header fields which are no longer of any use.

As in the previous embodiment, at each node Ni of the transmission path, the value Mi of each dedicated field Ci contained in the received frame T4 is decremented by one unit, in order to obtain new values ​​M'i. Then we proceed to the comparison of each new value M'i to zero (0). When a dedicated field Ci has a new value M'i null, we extract the packet Pi associated with it. On the other hand, when a field Ci has a new value M'i nonzero, we replace its value Mi (received) by the new value M'i corresponding. Then, the following node of the transmission path is transmitted to the frame T4 comprising the dedicated fields Ci which are associated with the packets remaining Pi and which comprise the new values ​​M'i. In other words, we preferentially remove from the frame T4 the dedicated field Ci whose new decremented value M'i is zero, before transmitting it to the next node. Of course, if we have to insert a new packet, possibly encapsulated, in the frame T4 to be transmitted, we can reuse the field Ci for this new packet, but with a new value Mi. Moreover, if several packets Pj, possibly encapsulated, must be inserted into the frame T4 to be transmitted, they are associated with as many new fields Cj as necessary.

The invention thus makes it possible to dispense not only with the G.709 headers, but also with the address tables necessary for the processing of the addresses for the packets in transit, in the networks of the prior art, since the tables of FIG. addressing are still needed in the UNI for inserting and retrieving packets. The processing at a node is therefore limited to reading the value Mi of each dedicated field Ci of the received frame, then the decrementation of a unit of each value Mi in order to obtain the new values ​​M'i, then the comparison of each M'i with the chosen comparison value (for example zero (0)), and finally the decision as to the extraction or the continuation of the transmission of each packet Pi, taking into account the result of the comparison of the new value M'i corresponding.

It is important to note that the control plane or the network management plan can be used to provide the nodes with the names of nodes that belong to the same subgroup (or domain) as theirs. In other words, one can dynamically vary the definitions of subgroups of nodes. Moreover, one can also dynamically vary the definitions of the different access buses of a network, that is to say the groups of nodes that are coupled to the different access buses, always by means of the control plane or the network management plan.

The invention is not limited to the embodiments of the processing module MT, network node Ni and transmission method described above, only by way of example, but it encompasses all the variants that can be considered by the skilled in the art within the scope of the claims below.