FR2883438A1 - Routing table constructing method for communication network, involves allocating transmission path to remaining secondary commutation units, where path comprises available links when streams are transported to one primary commutation unit - Google Patents

Abstract

The method involves allocating roots of N1 covering trees to primary commutation units, where the trees transport N1 streams simultaneously. A transmission path is allocated to each remaining secondary commutation units, where the path comprises a set of links that are available when the N1 streams are transported towards one of the primary commutation units. The path comprises a covering tree having the root for one primary commutation unit. Independent claims are also included for the following: (1) a method for transmission of a set of streams between N nodes through a commutation device (2) a computer program product comprising program code instructions for the execution of stages of a routing table constructing method (3) an information storage medium comprising instructions for a computing program adapted to implement a routing table constructing method (4) a commutation device for the transmission of a set of stream between nodes (5) a node for the transmission of a stream through a commutation device.

Description

Method of constructing a routing table and method of broadcasting,

  computer program product, storage means, switching device and corresponding node.

  FIELD OF THE DISCLOSURE The field of the invention is that of communication networks interconnecting a plurality of terminals.

  More specifically, the invention relates to a method of constructing a routing table and a method of simultaneously broadcasting a plurality of streams in a switched network, using the previously constructed routing table.

  The invention applies in particular, but not exclusively, to the simultaneous broadcasting of streams within a home audiovisual communication network.

  2. Prior Art US Patent 5,872,773 discloses a system and method for routing cells in a wireless communication network. The network comprises a plurality of switching nodes and the cells are routed according to particular trees, called VPI trees, which are rooted on the destination side and use Virtual Path Identifiers (VPIs). A routing protocol is provided to determine pre-established VPI trees rooted at each destination node. The routing protocol manages the paths of these trees, ensuring that there are at least two VPI trees from each source to each destination, for reasons of reliability, and that each destination node has multiple VPI trees for reasons of load distribution. The routing protocol includes an initial (off-line) procedure for determining the initial VPI trees. In order to take into account changes in network conditions and traffic, the routing protocol updates VPI tree paths dynamically and distributed. Updates are triggered by congestions, node failures or links, and additions of nodes or links.

  US Patent 6,400,681 discloses a packet switching communication network comprising a plurality of nodes interconnected by links. It is proposed to minimize the connection establishment time in the access nodes, and in particular the time for selecting an optimal path through the network, between an access node and a destination node. Each node of the network comprises: one or more communication adapters for receiving and transmitting data packets; a routing controller for allocating, controlling and managing network resources; a topological database, for storing network traffic and configuration characteristics, updated by means of network control messages; and a routing database, updated simultaneously with the topological database, and making it possible to store the computed or selected paths with their characteristics.

  It appears that none of the techniques known to date, including the two recalled above, are aimed specifically at maximizing the number of concurrent streams broadcast simultaneously within a network, each from a node distinct from that network. network.

  3. OBJECTIVES OF THE INVENTION The invention particularly aims to overcome these various disadvantages of the state of the art.

  More precisely, one of the objectives of the present invention is to provide a technique for simultaneously broadcasting a plurality of streams in a network, making it possible to maximize the number of concurrent concurrent broadcast streams.

  The invention also aims to provide such a technique not requiring dynamic bandwidth reservation mechanism, such a mechanism based on a signal that is generally relatively complex and clutter the network.

  Another object of the invention is to provide such a technique which is simple to implement and inexpensive.

  4. SUMMARY OF THE INVENTION These objectives, as well as others which will appear subsequently, are achieved according to the invention by means of a method of constructing a routing table, with a view to the broadcasting of a plurality of streams between N nodes through a switching device comprising N switching units. According to the invention, the method comprises: a first step of assigning to N1 switching units, called primary units, roots of N1 covering trees capable of carrying N1 streams simultaneously; a second step of assigning to each switching unit belonging to the remaining N-N1 units, called secondary units, a broadcasting path comprising in a first part one or more links, at least partially available when said N1 streams are transported simultaneously, to a primary unit, said associated primary unit, and in a second part the cover shaft rooted in said associated primary unit.

  The general principle of the invention therefore consists in forming one or more pairs of switching units each comprising a primary unit and a secondary unit, and sharing between the primary and secondary units of the same pair the cover shaft ( and therefore the corresponding bandwidth) whose root is assigned to the primary unit.

  An additional stream, broadcast from a secondary node, is a separate stream of the N1 streams that can be simultaneously simultaneously broadcast from one of the N1 primary units.

  A spanning tree is called a spanning tree to reach the set of nodes through the switching units (associated with the nodes) with a bandwidth equal to the capacity of the link between a source node and the first switching unit. which it is connected.

  Advantageously, the method comprises a preliminary step of determining the maximum number of recovery trees allowing a simultaneous transport of flows.

  Advantageously, the number N1 of primary units to which the roots of the cover trees are assigned is chosen equal to said maximum number.

  Preferably, the method comprises a step of choosing the N1 primary units and the associated secondary units in such a way that the number of links in the first part of the broadcast path of each secondary unit is minimal and that the secondary units are uniformly distributed on the associated primary units.

  In a particular embodiment, the second part of the broadcast path assigned to each secondary unit comprises the diminished cover shaft of said secondary unit if it is in an end of said shaft.

  The invention also relates to a method of broadcasting a plurality of streams between N nodes through a switching device comprising N switching units. According to the invention, this broadcasting method comprises the following steps: a) reading a routing table obtained by implementing the aforementioned method of constructing a routing table; b) broadcasting a stream from a source node through a switching unit of the switching device connected to that source node, using a broadcast path assigned to said switching unit in the routing table.

  Advantageously, when the switching unit is a secondary unit, step b) itself comprises the following step: insertion by the source node, said secondary node, of at least one broadcast information in the stream, said additional stream, indicating to the primary unit associated with said secondary unit that it must broadcast the additional stream using the recovery tree whose root has been assigned to it.

  Advantageously, the secondary node inserts an indicator in the header included in each of the packets of the supplementary stream, and when it detects said indicator, the associated primary unit sets, in a routing field included in each of the packets, a routing sequence representing the recovery tree whose root has been assigned to it.

  According to an advantageous variant, when the switching unit is a secondary unit, step b) itself comprises the following step: insertion by the source node, said secondary node, in the flow, said additional flow, of at least one routing information relating to the recovery tree whose root has been assigned to the primary unit associated with said secondary unit, so that the associated primary unit diffuses the additional stream using the recovery tree whose the root has been assigned to it.

  Advantageously, in the case of this variant, the secondary node places in a routing field, included in each of the packets of the additional stream, a routing sequence including a routing sub-sequence representing the recovery tree whose root has been assigned to the associated primary unit, so that the associated primary unit uses the routing subsequence to broadcast the packets along the collection tree whose root has been assigned to it.

  In a particular embodiment, step b) itself comprises the following steps: when it receives a first broadcast request from the source node connected to it, the secondary unit sends to the primary unit a second broadcast request indicating a bandwidth required for said additional stream; when it receives the second broadcast request, the primary unit determines whether it has at least the required bandwidth on the recovery tree whose root has been allocated to it; if it has at least said necessary bandwidth, the primary unit reserves said necessary bandwidth and informs the secondary unit. Advantageously, if it does not have at least said necessary bandwidth, the primary unit informs the secondary unit.

  In a particular embodiment, step b) itself comprises the following steps: when it receives a first broadcast stop request from the source node connected to it, the secondary unit sends a second request stopping broadcasting to the primary unit; when it receives the second broadcast stop request, the primary unit releases a bandwidth previously reserved for said additional stream.

  The invention also relates to a computer program product comprising program code instructions for executing the steps of the aforementioned method, when said program is executed on a computer.

  The invention also relates to an information storage means, possibly totally or partially removable, readable by a computer system, and comprising instructions for a computer program adapted to implement the aforementioned method, when said program is executed on a computer.

  The invention also relates to a switching device for broadcasting a plurality of streams between N nodes comprising N switching units, each of the switching units comprising a port of a first type for the connection of a node and at least a port of a second type for the connection of another switching unit.

  According to the invention, the device further comprises: first means for assigning to N1 switching units, called primary units, roots of N1 covering trees capable of carrying N1 streams simultaneously; and second means for assigning to each switching unit belonging to the remaining N-N1 units, called secondary units, a broadcasting path comprising in a first part one or more links, at least partially available when said N1 streams are transported. simultaneously, to a primary unit, said associated primary unit, and in a second part the cover shaft rooted in said associated primary unit.

  The invention also relates to a node for broadcasting a stream through a switching device comprising N switching units, said node comprising a port of a first type for connecting the node to a switching unit of the switching device. . According to the invention, the node further comprises: first means for assigning to N1 switching units, called primary units, roots of N1 covering trees capable of carrying N1 streams simultaneously; and second means for assigning to each switching unit belonging to the remaining N-N1 units, called secondary units, a broadcasting path comprising in a first part one or more links, at least partially available when said N1 streams are transported. simultaneously, to a primary unit, said associated primary unit, and in a second part the cover shaft rooted in said associated primary unit.

  5. List of Figures Other features and advantages of the invention will appear on reading the following description of a preferred embodiment of the invention, given by way of indicative and non-limiting example, and the accompanying drawings. in which: Figure la is a flowchart of a particular embodiment of the method of constructing a routing table according to the invention; FIG. 1b is a flowchart of a particular embodiment of the method of simultaneous diffusion of a plurality of streams according to the invention; FIG. 2 represents a home communication network in which the invention can be implemented in a preferential manner; FIG. 3 represents the partial topology of the central switch appearing in FIG. 2 and in which six primary nodes (and hence six primary switching units) can be selected according to the invention; FIGS. 4 to 9 illustrate, in the context of the exemplary topology of FIG. 3, six cases of diffusion of a stream from each of one of the six primary nodes (and therefore of one of the six possible primary switching units), as well as the cumulative relative loads on the X-type links; FIG. 10 illustrates, in the context of the exemplary topology of FIG. 3, the association of six secondary nodes with the six primary nodes (and thus of six secondary switching units with the six primary switching units), according to FIG. invention; FIG. 11 presents a flowchart of a systematic search algorithm of all the configurations allowing the implementation of the invention, in the context of the example of topology of FIG. 3; FIG. 12 illustrates an exemplary bandwidth sharing protocol between a primary node and the associated secondary node, according to the invention; FIG. 13 shows an exemplary algorithm performed by a secondary node for implementing the exemplary protocol illustrated in FIG. 12; and FIG. 14 shows an example of an algorithm performed by a primary node to implement the protocol example illustrated in FIG. 12. 6. Detailed description We now present, in connection with the flowchart of FIG. main steps of the method according to the invention for constructing a routing table In step E201, the maximum number N1 of recovery trees is determined, allowing a simultaneous transport of flows. This determination depends on the architecture of the switching device and the capabilities of the links connecting the switching units to each other and to the nodes.

  In step E202, one chooses N switching units, among the N, to which will be assigned the root of a recovery tree (primary units). This assignment is performed in step E203. The remaining units, when Ni <N, are secondary units that are associated with primary units for streaming.

  In step E204, each secondary unit is assigned a broadcast path which comprises, in a first part, one or more links, at least partially available when said N1 streams are transported simultaneously, to the primary unit associated with it, and in a second part the recovery tree rooted in this associated primary unit.

  In step E205, the routing path assigned to it is saved in a routing table for each switching unit (or node). This path is equal to or includes a cover shaft depending on whether the unit is primary or secondary. This routing table can be saved at a node or in the switching device according to the embodiments.

  A particular embodiment of the method for the simultaneous diffusion of a plurality of streams according to the invention is now presented with reference to FIG. In a step E101, the previously saved routing table is read (see step E205 of FIG. In a step E102, for the broadcast of an additional stream from a source node through the secondary unit connected to it, there is use of the affected collection tree, in the routing table, to the associated primary unit.

  FIG. 2 represents a home communication network in which the invention can be implemented in a preferential but nonlimiting manner.

  This network comprises N nodes, referenced 101a, 101b, 101c, 101d, 101e, 101f, 101g, 101h, 101i, 101j, 101k and 1011, representing multimedia interface equipment. Each node is typically placed in a room of the house and thus allows to connect the terminals in this room to other terminals in other parts of the house. The nodes are connected to a central switch 100 through links referenced 102a, 102b, 102c, 102d, 102e, 102f, 102g, 102h, 102i, 102j, 102k and 1021. These links are for example of the UTP5 type (of the English Unshielded Twisted Pair, category 5) as specified in the ANSI / TIA / EIA / 568A standard conventionally used in Ethernet type networks. It should be noted that other types of links can be used such as fiber links.

  The nodes as well as the central switch constitute a federating network allowing the exchange of data between terminals in the dwelling.

  The central switch 100 consists of N switching units (or ESUs, for Elementary Switching Units), referenced 10, 11, 12, 13, 20, 21, 22, 23, 30, 31, 32, 33, allowing the routing of data between the nodes. In this example, the central switch 100 comprises twelve switching units (N = 12). Each switching unit has, in this example, three ports of a first type called X to connect the switching units to each other, and a port of a second type called Y to connect a switching unit to a node.

  In the remainder of the description, a primary switching unit is a switching unit connected to a primary node by a link Y, and a secondary switching unit a switching unit connected to a secondary node by a link Y. In the mode of embodiment shown in Figure 2, the type X ports are connected by internal links to the switch 200 (also called X links), and the type Y ports are connected to the nodes by external links (also called Y links). Both types of links may have different characteristics, such as the maximum data rate they can carry.

  In an alternative embodiment of the central switch 100, the switching units are distributed at the nodes. The links connecting the X ports thus become external links and the links connecting the Y ports to the internal link nodes. The number of switching units and the way in which they are connected may also vary.

  For the sake of simplification, no terminal has been shown in Figure 2. However, it is understood that different types of terminals can be connected to the backbone. For example, a multimedia decoder (set top box in English language) could be connected to the node 101a, while a digital camera would be connected to the node 101e and televisions would be connected respectively to the nodes 101c, 101f and 101h.

  In general, in such a home communication network, the number N1 of primary nodes is less than or equal to a maximum number B given by the following equation: B = p * c + int [(p * c) / (N-1)] (1) where: - the function int Q is the function integer value; - the parameter c is the ratio between the capacity of an X link and the capacity of a link Y. For example, if the capacity of a link Y is 100 Mbit / s and the capacity of an X link is 200 Mbit / s, then we have: c = 2. If a part of the bandwidth is reserved for the asynchronous traffic, only the bandwidth relative to the synchronous traffic is taken into account to determine the parameter c; the parameter p is the number of X type ports included in each switching unit.

  The streams to be broadcast are injected into the network through a link Y, so their bandwidth is limited to the capacity of a link Y. Furthermore, if it is assumed that each link X comprises two unidirectional links of opposite meanings, then the number of unidirectional links at least partially free, when B primary nodes are active at the same time and with the complete bit rate of each Y-type link, is less than or equal to a maximum number R given by the following equation: R = p * c (N-1) * int [(p * c) / (N-1)] (2) Equations (1) and (2) above are true only if the network topology does not have a bottleneck. throttle.

  If the following condition is satisfied: p * c <N-1, then equations (1) and (2) are simplified to: B = R = p * c.

  For example, if we assume that: p = 3, C = 2 and N = 12 (which corresponds to the case of figure 2), then B = R = 6. This means that, in this example, it is possible to anytime to broadcast six streams from the six primary nodes (simultaneously and with full bandwidth of link Y), while having six unidirectional links that remain half free.

  As explained in detail below, the present invention makes it possible to ensure, in such a network context, that additional streams are broadcast from the remaining nodes (secondary nodes), thanks to shared bandwidth and trees. overlay allocated to the primary nodes.

  The partial topology of the central switch 100 appearing in FIG. 2 and in which six primary nodes (and hence six primary switching units) can be selected according to the invention are now presented in connection with FIG.

  This FIG. 3 details only the interconnection of the switching units 10, 11, 12, 13, 20, 21, 22, 23, 30, 31, 32, 33 via the bidirectional X links. As a result, the Y links are not shown in Figure 3. In this example, twelve switching units are interconnected by eighteen bidirectional X links.

  Figure 3 also shows the numbering of the three X-type ports of each switching unit (number 1, 2 or 3 placed in a rectangular frame near the relevant node and X link).

  The route of a data packet from a given node is represented by a sequence of X-type port numbers, i.e. one per traversing switching unit. This sequence is the routing header of the packet. For example, if a packet is derived from the link Y connected to the referenced switching unit 10 and includes the routing header 32313132313, then the referenced switching unit 10 transmits this packet through its port 3 to the exchange unit. switching referenced 32, the referenced switching unit 32 retransmits this packet through its port 2 to the referenced switching unit 31, the referenced switching unit 31 retransmits this packet through its port 3 to the referenced switching unit 13, the referenced switching unit 13 retransmits this packet through its port 1 to the referenced switching unit 12, and so on; finally, the switching units are traversed in the following order: 10-32-31-13-12-20-21-33-30-22-23-11.

  It is recalled that, in general, the packets can be of point-to-point type (a source, a destination) or point-to-multipoint (a source, several destinations). Broadcast packets are a special case of the second case (they are sent from one source to all other nodes in the network). Broadcast packets are routed along a spanning tree (for other point-to-multipoint packets, it may be sufficient to use a portion of a spanning tree).

  As explained in the French patent application FR 2 824 434, a cover shaft is preferably linear, which is always possible to obtain with the present network topology. The route of such linear overlay trees can be described analogously to the point-to-point packet route in the form of a sequence of X-type port numbers to which the packet is sent by the successive switching units. . The length of a road of a recovery tree is equal to the number of switching units traversed after the packet has been introduced by one of the switching units, that is to say the number of nodes minus one. This length is therefore equal to 11 in the above example where the central switch comprises 12 switching units.

  By performing an exhaustive search (with a computer program) among all 11 theoretically possible routes (3 "= 177,147 routes), the inventors have found that the following 38 roads (and only they) represent tree cover allowing diffusion in the context shown in Figure 3: 12131312131 12132123121 12132123212 12132131213 13121313121 13121313212 13131213131 13131313131 13132123131 13132313132 13212321231 13231313231 13231321232 13231321323 21231312131 21232123121 21232123212 21232131213 21313121313 23131323131 23212313231 23212321232 23212321323 23213132313 31213123121 31213123212 31213131213 31232123213 31312131312 31312321313 31313131313 31313231313 31323131232 31323131323 32312313231 32312321232 32312321323 32313132313 is note that due to the internal symmetry of the network (which is viewed in the same way from any node / switching unit) and the way in which X-type ports are digitized the representation of a spanning tree does not depend on the switching unit at the root of the spanning tree. This facilitates the implementation of the routing scheme.

  FIG. 4 illustrates, in the context of the topology example of FIG. 3, the broadcasting of a stream from the primary switching unit referenced 10 and using the routing header 32313132313. The stream reaches the switching units (and thus the corresponding nodes) in the following order: 10-32-31-13-12-20-21-33-30-22-23-11.

  The triangular arrows near each X link (one arrow for each direction, ie for each of the two unidirectional links constituting the X link) indicate the link capacities used by this stream, in Y link capacity unit. (in this example 100 Mbitls). It is recalled that the bandwidth of each individual stream is limited by this link capacity Y. Note also that the cumulative bandwidth of all the isochronous flows occupying a link Y may not reach the limit of 100 Mbit / s, since part of the IEEE 1394 bandwidth, typically 20%, is reserved for asynchronous traffic. However, this does not change the relationship between the capabilities of the links (that is, an X link for example always has twice the capacity of a Y link).

  Note also that X links like Y links are of type fullduplex, which means that they can use their entire bandwidth independently in each direction. This is why two triangular arrows appear near each X-link, one for each direction.

  FIG. 5 illustrates, in the context of the topology example of FIG. 3, the broadcasting of a stream from the primary switching unit referenced 13 and using the routing header 32313132313. The stream reaches the switching units (and thus the corresponding nodes) in the following order: 13-31-32-10-11-23-22-30-33-21-20-12. Here, the triangular arrows indicate for each link X and each direction, the capacity used by this flow as well as by the flow described above in relation with FIG. 4.

  FIG. 6 illustrates, in the context of the topology example of FIG. 3, the broadcasting of a stream from the primary switching unit referenced 20 and using the routing header 23212321232. The stream reaches the switching units (and thus the corresponding nodes) in the following order: 20-23-11-12-13-10-32-31-30-33-21-22. Here, the triangular arrows indicate for each link X and each direction, the capacity used by this stream as well as by the flows described above in relation to FIGS. 4 and 5.

  FIG. 7 illustrates, in the context of the topology example of FIG. 3, the broadcasting of a stream from the primary switching unit referenced 23 and using the routing header 23212321232. The stream reaches the switching units (and thus the corresponding nodes) in the following order: 23-20-12-11-10-13-31-32-33-30-22-21. Here, the triangular arrows indicate for each link X and each direction, the capacity used by this stream as well as by the flows described above in relation to FIGS. 4 to 6.

  FIG. 8 illustrates, in the context of the topology example of FIG. 3, the broadcasting of a stream from the primary switching unit referenced 30 and using the routing header 13212321231. The stream reaches the switching units (and thus the corresponding nodes) in the following order: 30-31-13-10-11-12-20-23-22-21-33-32. Here, the triangular arrows indicate for each link X and each direction, the capacity used by this flow as well as by the flows described above in connection with FIGS. 4 to 7.

  FIG. 9 illustrates, in the context of the topology example of FIG. 3, the broadcasting of a stream from the primary switching unit referenced 33 and using the routing header 13212321231. The stream reaches the switching units (and thus the corresponding nodes) in the following order: 33-32-10-13-12-11-23-20-21-22-30-31. Here, the triangular arrows indicate for each link X and each direction, the capacity used by this flow as well as by the flows described above in connection with FIGS. 4 to 8.

  As can be seen in this figure 9, of the 36 half-links X (ie the 36 unidirectional links) of the network, most are fully occupied by two flows.

  Only the following six half-links X have half of their total capacity which remains free: from the switching unit referenced 11, port 1, to the primary switching unit referenced 10, port 1; from the switching unit referenced 12, port 1, to the primary switching unit referenced 13, port 1; from the referenced switching unit 21, port 1, to the primary switching unit referenced 20, port 1; from the switching unit referenced 22, port 1, to the primary switching unit referenced 23, port 1; from the switching unit referenced 31, port 1, to the primary switching unit referenced 30, port 1; from the referenced switching unit 32, port 1, to the primary switching unit referenced 33, port 1.

  FIG. 10 illustrates the association (in pairs) of six secondary nodes to the six primary nodes, i.e. the association of six secondary switching units 11, 12, 21, 22, 31, 32 to the six units. primary switching circuits 10, 13, 20, 23, 30, 33 presented above in connection with FIGS. 4 to 9.

  To realize these paired associations (10, 11), (13, 12), (20, 21), (23, 22), (30, 31) and (33, 32), the 6 half-links X are used. mentioned above which are half free. Thus, by using the bandwidth that remains available on these 6 half-links X, the secondary switching units (and therefore the secondary nodes) are allowed to share the bandwidth allocated (statically) to the primary switching units (and thus to the primary nodes).

  For example, the secondary node connected to the secondary switching unit referenced 11 can transmit a flow at 80 Mbit / s, which passes through the secondary switching unit referenced 11 and is transmitted to the primary switching unit referenced 10 to be broadcast on the network, while at the same time the primary switching unit referenced 10 also broadcasts another 80 Mbit / s stream injected on its type Y port by the primary node which is connected thereto.

  Bandwidth sharing between a primary node and its associated secondary node can be done in a number of ways. Two solutions are described below through the following example: the secondary switching unit referenced 11 must broadcast a stream using the primary switching unit referenced 10 and the recovery tree assigned to the latter.

  According to the first solution, the secondary switching unit referenced 11 sends isochronous flow packets to the primary switching unit referenced 10, using a point-to-point routing header 1 and positioning a special bit in the package header. The primary switching unit referenced 10 transmits the packets to the primary node to which it is connected (via a link Y). This primary node, when it detects the special bit placed in the header of the packets, transmits the packets on an IEEE 1394 bus connected to it, and furthermore sends the packets to the primary switching unit referenced 10. last retransmits the packets using the broadcast header assigned to it (32313132313).

  According to the second solution, the secondary switching unit referenced 11 prefixes the broadcast header (32313132313) assigned to the primary switching unit referenced 10, with the port number 1 used to go from the switching unit. secondary referenced 11 to the primary switching unit referenced 10. There is thus obtained the following completed broadcast header: 132313132313. Sending, from the secondary switching unit referenced 11 to the primary switching unit referenced 10, packets with this completed broadcast header causes the crossing of the switching units (and therefore the corresponding nodes) in the following order: 11-10-32-31-13-12-20-21-33-30-22-23-11.

  Optionally, in this particular case where the secondary switching unit referenced 11 reappears, superfluously, in the last position of the broadcast route (and more generally in all cases where the secondary secondary switching unit reappears in the last position of the broadcast route), we delete the last hop of the routing header. In the aforementioned example and with the second solution described above, the following broadcast header is obtained: 1323 13 1323 1. This can also be applied with the first solution described above.

  With the exemplary topology illustrated in FIGS. 3 to 10, this optimization option is also possible for the pair of switching units (13, 12). On the other hand, it can not be applied for the other four pairs of switching units (20, 21), (23, 22), (30, 31) and (33, 32) since the routing paths used do not end at the starting node.

  While FIGS. 4 to 10 were intended to present the principle of the invention through a particular flow routing scheme, FIG. 11 shows a systematic search algorithm of all the configurations allowing the implementation of the invention, in the framework of the topology example of Figure 3.

  This algorithm consists of a systematic search: on the one hand among the 924 possible selections of the 6 primary nodes among the 12 nodes (that is to say the possible selections of the 6 primary switching units out of the 12 switching units), and on the other hand, among the 386 = 3 010 936 384 possible combinations of 6 broadcast routing headers out of the 38 broadcast routing headers mentioned above in the discussion of Figure 3.

  12 12! i Note that: 924 = = 6 6! (12 6)! This gives a total of about 2.78 * 1012 combinations to examine. The algorithm can be modified slightly so that an incomplete combination (less than 6) of routing headers is skipped as soon as the load on certain link (s) exceeds 2 (ie say twice the capacity of a link Y, equal to 100 Mbit / s in the example above). This considerably reduces the number of combinations that are actually examined, and thus reduces the execution time of the algorithm. This modification, which does not appear in FIG. 11, does not change the principle of the algorithm.

  With the example of topology of FIG. 3, the result of the execution of the search algorithm indicates that 414 of the 924 possible selections of 6 primary nodes out of 12 make it possible to implement the present invention, with a total of 13552 possible combinations of 6 broadcast routing headers out of 38 (between 7 and 168 combinations for each selection of 6 primary nodes).

  This search algorithm can be described by the following pseudo code: For each subset S of 6 elements of set N of 12 nodes For each combination of 6 out of 38 broadcast routing headers Set to 0 of all link loads For each of the 6 nodes of subset S and its associated routing header Determination of the route through the network Incrementation of the load of each of the 11 links involved If all the loads of the paths are less than or equal to 2 gives the empty value to the variable set_subsc_codes The empty set is given to the variable set_computing_node For all links with a load equal to 1 Adding the source node of the link to the variable set_subsc_code Adding the destination node of the link to the set setpoint_nceuds_destination If set_neups_source = S and set_count_nodes = N \ S (where N \ S 15 denotes the difference of sets N and S) Retention of the combination of 6 nodes and 6 en-tê With the aid of FIGS. 12, 13 and 14, an exemplary bandwidth sharing protocol between a primary node and the associated secondary node 20 according to the invention is now presented.

  This protocol is based on the fact that the primary nodes manage the bandwidth occupied on the one hand by the flows of which they are the origin and on the other hand by the flows coming from the associated secondary nodes. Such bandwidth management methods are well known to those skilled in the art and are therefore not described further.

  detail in the present description.

  Three scenarios, A, B and C are respectively presented successively in FIG. 12, respectively corresponding to a successful connection attempt, a disconnection and a failed connection attempt.

  In a first scenario (denoted by A in FIG. 12), the bandwidth sharing protocol is such that: when it receives a first broadcast request (CONNECTION_REQUEST) from a terminal connected to it (for example more to a JOIN message, as described in the IEEE 1394.1 standard, sent to the talking portal) (step 131, fig.13), the secondary node determines the bandwidth necessary to satisfy the first request (step 132, fig.13). Then, the secondary node sends to its associated primary node a second broadcast request 121 indicating in particular said necessary bandwidth (step 133, fig.13); when it receives the second broadcast request (step 141, FIG. 14), the primary node determines whether it has at least said necessary bandwidth on the cover tree assigned to it (step 142, fig.14). It thus processes this second broadcast request as it processes requests from peripheral terminals that are directly connected to it. Indeed, the streams coming from these peripheral terminals actually share the broadcast route (overlay tree) with the streams coming from the secondary node in question; if it has at least the necessary bandwidth (affirmative answer to the question of step 143, fig. 14), the primary node reserves it and informs the secondary node by sending it a positive acknowledgment message 122 ( step 144, Fig. 14); the secondary node, which until now had waited for an acknowledgment message (step 133, FIG. 13), receives the positive acknowledgment message 122 (affirmative answer to the question of step 135, FIG. begin transmitting the stream using the predetermined route comprising as said first intermediate node said primary node (step 136, fig.13). Depending on the high-level protocol used, other actions may be performed (for example, sending a STREAM_STATUS message, in accordance with IEEE 1394.1, from the source gate to the destination gate).

  In a second scenario (denoted B in FIG. 12), the bandwidth sharing protocol is such that: when the peripheral terminal connected to the secondary node stops transmitting the stream (for example, following the sending of a LEAVE message, as described in the IEEE 1394.1 standard) (step 137, FIG. 13), the secondary node stops the transmission of the stream (step 138, FIG. 13) and sends a broadcast stop request (DISCONNECTION_REQUEST) 123 at the associated primary node (step 139, fig.13); when it receives the broadcast stop request (step 145, FIG. 14), the primary node releases the necessary bandwidth that was occupied by the interrupted stream coming from the secondary node (step 146, FIG. would do for a stream coming from a peripheral terminal directly connected to it. In addition, high-level protocol layers may be informed (for example, by sending a STREAM_STATUS message, in accordance with IEEE 1394.1, from the source gate to the destination gate).

  In a third scenario (denoted C in FIG. 12), the bandwidth sharing protocol is such that: - if after having received the second broadcast request 121 (see the description of the first scenario above), the primary node determines that it does not have the necessary bandwidth (negative answer to the question of step 143, Fig. 14), then it informs the secondary node by sending a negative acknowledgment message 124 (step 147, fig.14); when it receives the negative acknowledgment message 124 (negative answer to the question of step 135, FIG. 13), the secondary node then considers that the flow connection establishment is a failure (step 1310, FIG. 13). Higher level notification may be required by relevant high-level protocol layers (for example, by sending a STREAM_STATUS message, in accordance with IEEE 1394.1, from the source gate to the destination gate).

  The required bandwidth is not available if the overlay tree (broadcast route) is already busy with flows from peripheral terminals connected to the primary or secondary node.

Claims (21)

  A method of constructing a routing table, for the purpose of broadcasting a plurality of streams between N nodes through a switching device comprising N switching units, said method is characterized in that it comprises: a first step (E203) of assignment to N1 switching units, said primary units, roots of N1 covering trees that can carry N1 streams simultaneously; a second step (204) of assignment to each switching unit belonging to the remaining N-N1 units, called secondary units, a broadcasting path comprising in a first part one or more links, at least partially available when said N1 flows are transported simultaneously, to a primary unit, said associated primary unit, and in a second part the cover shaft rooted in said associated primary unit.   2. Construction method according to claim 1, characterized in that it comprises a preliminary step (E201) for determining the maximum number of covering shafts for simultaneous transport of flows.   3. Construction method according to claim 2, characterized in that the number N1 of primary units to which the roots of the covering shafts are assigned is chosen equal to said maximum number.   4. Construction method according to claim 1, characterized in that it comprises a step (E202) of choice of N1 primary units and associated secondary units performed in such a way that the number of links in the first part of the path of each secondary unit is minimal and that the secondary units are uniformly distributed on the associated primary units.   5. Construction method according to claim 1, characterized in that the second part of the broadcast path assigned to each secondary unit comprises the reduced cover shaft of said secondary unit if it is in an end of said shaft.   A method of broadcasting a plurality of streams between N nodes through a switching device comprising N switching units, said method is characterized in that it comprises the following steps: a) reading (E101) a table routing obtained by implementing the method of constructing a routing table according to any one of claims 1 to 5; b) broadcasting (E102) a stream from a source node through a switching unit of the switching device connected to that source node, using a broadcast path assigned to said switching unit in the routing table .   7. broadcast method according to claim 6, characterized in that, when the switching unit is a secondary unit, step b) itself comprises the following step: - insertion by the source node, said secondary node at least one broadcast information in the stream, said supplementary stream, indicating to the primary unit associated with said secondary unit that it must broadcast the additional stream using the recovery tree whose root has been assigned to it .   Broadcasting method according to claim 7, characterized in that the secondary node inserts an indicator in the header included in each of the packets of the supplementary stream, and in that, when it detects said indicator, the primary unit associated, in a routing field included in each packet, a routing sequence representing the recovery tree whose root has been assigned to it.   9. broadcast method according to claim 6, characterized in that, when the switching unit is a secondary unit, step b) itself comprises the following step: - insertion by the source node, said secondary node in the stream, said additional stream, at least one routing information relating to the recovery tree whose root has been assigned to the primary unit associated with said secondary unit, so that the associated primary unit diffuses the extra flow using the overlay tree whose root has been assigned to it.   10. Broadcast method according to claim 9, characterized in that the secondary node places in a routing field, included in each of the additional stream packets, a routing sequence including a routing sub-sequence representing the recovery tree. whose root has been assigned to the associated primary unit, so that the associated primary unit uses the routing subsequence to broadcast the packets along the collection tree whose root has been assigned to it. 11. Broadcast method according to claim 7, wherein step b) itself comprises the following steps: when it receives a first broadcast request from the source node connected to it ), the secondary unit sends to the primary unit a second broadcast request (121) indicating a bandwidth required for said additional stream (133); when it receives the second broadcast request (141), the primary unit determines whether it has at least said necessary bandwidth (142) on the collection tree whose root has been allocated to it; if it has at least said necessary bandwidth, the primary unit reserves said necessary bandwidth (142) and informs the secondary unit (122; 144) thereof.   12. Broadcasting method according to claim 11, characterized in that, if it does not have at least said necessary bandwidth, the primary unit informs the secondary unit (122; 147).   13. Broadcasting method according to any one of claims 7 to 12, characterized in that step b) itself comprises the following steps: - when it receives a first broadcast stop request from the source node which connected to it (137), the secondary unit sends a second broadcast stop request (123) to the primary unit (139); when it receives the second broadcast stop request (145), the primary unit releases a bandwidth previously reserved for said additional stream (146).   14. Computer program product characterized in that it comprises program code instructions for executing the steps of the method according to any one of claims 1 to 13, when said program is executed on a computer.   15. An information storage medium, possibly totally or partially removable, readable by a computer system, characterized in that it comprises instructions for a computer program adapted to implement the method according to any one of claims 1 to 13, when said program is run on a computer.   A switching device for broadcasting a plurality of streams between N nodes comprising: N switching units, each of the switching units comprising a port of a first type for the connection of a node and at least one port of a second type for connecting another switching unit; the device is characterized in that it further comprises: first allocation means to N1 switching units, called primary units, roots of N1 covering trees capable of carrying N1 streams simultaneously; and second means for assigning to each switching unit belonging to the remaining N-N1 units, called secondary units, a broadcasting path comprising in a first part one or more links, at least partially available when said N1 flows are transported simultaneously, to a primary unit, said associated primary unit, and in a second part the cover shaft rooted said associated primary unit.   17. Switching device according to claim 16, characterized in that it comprises means for determining the maximum number of cover shafts for simultaneous flow transport.   18. Switching device according to claim 16, characterized in that it comprises a routing table in which are saved information representative of the assignments of the recovery trees to the primary units and assignments of the broadcast paths to the secondary units.   19. Node for broadcasting a stream through a switching device comprising N switching units, said node comprising: a port of a first type for connecting the node to a switching unit of the switching device; the node is characterized in that it further comprises: first means for assigning to N1 switching units, called primary units, roots of N1 covering trees that can carry Ni streams simultaneously; and second means for assigning to each switching unit belonging to the remaining N-N1 units, called secondary units, a broadcasting path comprising in a first part one or more links, at least partially available when said N1 flows are transported simultaneously, to a primary unit, said associated primary unit, and in a second part the cover shaft rooted said associated primary unit.   20. Node according to claim 19, characterized in that it comprises means for determining the maximum number of cover shafts for simultaneous transport of flows.   21. Node according to claim 19, characterized in that it comprises a routing table in which are saved information representative of the assignments of the recovery trees to the primary units and assignments of the broadcast paths to the secondary units.