JP4894013B2 - System and method for searching a path between nodes included in a network - Google Patents

Description

  The present invention relates to various route searches in a network, and more particularly to disjoint route searches.

With the explosive development and spread of the Internet, computer networks are frequently used in work and daily life. In a computer network, a routing table is created based on information exchanged by a routing protocol, and packets are delivered according to the routing table. Typical routing protocols include OSPF (Open Shortest Path First), RIP (Routing Information Protocol), and SI-SI (Intermediate System to Intermediate System) used in a single autonomous system (AS (Autonomous System)). is there. In addition, there is a DSR (Dynamic Source Routing) protocol that has been proposed for a network (multi-hop wireless network, ad hoc network) using P2P and wireless (wireless) terminals. Furthermore, there is BGP (Border Gateway Protocol) which is a routing protocol between different ASs.
JP 2007-159099 A JP 2006-285386 A

  These routing protocols basically obtain network information (network topology) from a source (source node) to a destination node (destination node) by exchanging information between nodes (typically routers). A function and a function of determining an appropriate route between the source node and the destination node by an appropriate algorithm based on the acquired network information are provided. For example, in OSPF, a network topology database is constructed, and a routing table is created using a shortest path search algorithm called the Dijkstra method. In searching for the shortest route, it is desirable to consider not only the number of hops but also factors including the cost, bandwidth, and delay of the route (routing metric).

  Packet routing in an IP network typified by the Internet is performed by referring to a routing table held by the router in a passing router. In the routing table, the NHR (Next Hop Router (next hop)) corresponding to the destination network address is recorded, and the packet is transferred to the NHR corresponding to the network address that is the longest match with the destination address of each packet. Each router repeats the above operation to transfer the packet to the destination.

  In the future, as the scale of the network increases with the further spread of the Internet, it is considered that the burden of route search processing will increase. Moreover, the chance of searching for a new route increases due to a change in the state of the node. Typically, mobile terminals may be required to update the routing table more frequently as they join the Internet.

  Furthermore, in recent years, in an optical network, it has been proposed to consider not only the link speed but also a plurality of factors such as the success rate of wavelength reservation as the link cost. Further, since the link cost varies depending on factors such as the number of users using the link and the type of data passing through the link, it is desirable to update the routing table in consideration of the variation of those factors. Therefore, it is considered that the frequency of performing route search and updating the routing table further increases.

  Patent Document 1 describes a method of finding disjoint routes from a certain mobile node to a plurality of service nodes that can provide services to the mobile node in an ad hoc network including a plurality of nodes. The method includes the steps of transmitting a route request from a mobile node, distributing the route request by a node of the network, and tracing a route request from the mobile node to the service node when a certain route request is reached. Returning the included route response, determining the disjoint route based on the route response received by the mobile node, and the route response received by the mobile node corresponding to a sufficient number of disjoint routes. If not, issuing a new route request with an incremented path extension parameter.

  It is described that a disjoint route from a mobile route to a service node (gateway node) for connecting to the Internet is found. It describes that when connecting an ad hoc network to the Internet, if the route to the currently communicating gateway is disconnected, high-speed multi-hop to another gateway is performed to improve communication reliability. ing. In order to cope with such a situation, in order to always secure a plurality of routes, it is necessary to frequently perform a route search process, and it is also necessary to shorten a processing time required for the route search.

  Patent Document 2 describes an integrated circuit device of a type that dynamically reconfigures various data paths by changing the functions of a plurality of types of processing elements (PE) and their connection.

One embodiment of the present invention is a system for searching for a route from a source node to one or more destination nodes in a network including a plurality of nodes and a plurality of links connecting the nodes. The system includes a plurality of virtual node blocks respectively corresponding to a plurality of nodes included in the network, a plurality of virtual links respectively corresponding to a plurality of links, based on network information collected by one or a plurality of routing protocols. And a unit (virtual network configuration unit) that constitutes the virtual network that realizes the virtual network including the processor.

In this system, the processor may include a reconstruction unit. The reconfiguration unit includes a plurality of processing elements and a connection matrix capable of dynamically changing the connection of the plurality of processing elements, and is a unit capable of dynamically reconfiguring a circuit. The virtual network configuration unit configures a virtual network in the reconfiguration unit based on network information collected by one or more routing protocols. Furthermore, the units constituting the virtual network include a plurality of virtual node blocks respectively corresponding to a plurality of nodes included in the network by a plurality of processing elements, and a plurality of virtual nodes respectively corresponding to a plurality of links by a connection matrix. Configure the link.

  The units constituting the virtual network further configure the virtual source node block corresponding to the source node among the plurality of virtual node blocks so as to transmit search packets to all virtual links connected to the virtual source node block. . The unit constituting the virtual network further adds progress information including information on the virtual link and / or the virtual node block through which the received search packet has passed to each received search packet for each of the plurality of virtual node blocks. A search packet obtained by duplicating the received search packet is transmitted to all the virtual links connected to the respective virtual node blocks except for the virtual link where the received search packet has arrived.

  The system further determines from the source node the progress information of a plurality of search packets arriving at the virtual destination node block (virtual destination node block) corresponding to the destination node (destination node) among the plurality of virtual node blocks. A route selection unit is provided that selects a route that meets a predetermined condition from a plurality of routes reaching the destination node.

  The units constituting the virtual network and / or the path selection unit may be realized by a circuit configured in the reconfiguration unit, or may be realized by a processor such as RISC for controlling the reconfiguration unit. You may implement | achieve by the external processor physically different from the chip | tip containing a structural unit.

  In this system, a virtual network having a function of transferring a search packet is reconfigured in a reconfiguration unit that can dynamically reconfigure a circuit. Since the virtual node block configured in the reconfiguration unit only needs to have a function of copying and transferring the search packet under a predetermined condition, each virtual node block is implemented with one or a small number of processing elements (PE). it can. For this reason, a virtual network corresponding to a network (target network) to be searched for a route can be reconfigured in a reconfiguration unit having an appropriate number of PEs as hardware resources. Also, if the hardware resources of the reconfigurable unit that is made into chips are insufficient, multiple chips can be connected, or the virtual network corresponding to the target network can be multi-contexted and reconfigured in a time-division manner into reconfigurable units Is possible.

  And since the circuit can be dynamically reconfigured in the reconfiguration unit, when the information of the network (original network or target network) collected by the routing protocol is updated, it is updated to the reconfiguration unit based on it. Virtual network based on the information can be reconfigured.

  In this system, the virtual network can be reconfigured in the reconfiguration unit. For this reason, a route search can be executed by flowing a search packet through a virtual network without using a route search method such as the Dijkstra method that requires a lot of calculation time. In a virtual network configured in the reconfiguration unit, the search packet is transferred at the operation speed (clock speed) of the reconfiguration unit. Therefore, a route can be found in a short time compared with the method shown in Patent Document 1.

  Further, in this system, the search packet can be processed in the reconfiguration unit in which the virtual network is configured. Therefore, the route can be selected by the search packet received by the virtual destination node, and there is no need to return a response to the virtual source node. For this reason, it is possible to omit a time for waiting for a route response from another node, which occurs when a search packet is transmitted to the original network.

  The progress information preferably includes bitmap data that can identify a virtual link and / or a virtual node block that has been passed by a bit position. The route selection unit can select a disjoint route from a plurality of routes from the source node to the destination node by calculating a logical product of bit data of a plurality of pieces of progress information bitmap data.

  It is desirable that the plurality of processing elements of the reconstruction unit include a plurality of delay elements that can dynamically set the delay amount. A unit constituting the virtual network can form a plurality of virtual links by the connection matrix and a plurality of delay elements. By setting the delay amount corresponding to the cost of the corresponding link in the delay element included in each virtual link, the route selection unit can determine the route through which the search packet has passed according to the timing received by the virtual destination node block. Information including the link cost can be obtained, and the information can be used for route selection.

  Preferably, the route selection unit selects a route that meets a predetermined condition from a plurality of routes from the source node to the destination node, based on the progress information of the plurality of search packets in the early order of arrival at the virtual destination node block. In this system, since a search packet is transmitted to the virtual network, the order in which it arrives at the virtual destination node block (when it arrives first), the number of hops of the route in the target network, and other metrics are reflected. is there. Therefore, the route can be filtered using the information. Furthermore, the processing time required for route selection can be shortened.

  The units that make up the virtual network are configured so that each virtual node block is copied only from a plurality of search packets in the order of arrival in each virtual node block, except for the virtual link from which the received search packet has arrived. It is desirable to reconfigure to transmit to all virtual links connected to the virtual node block. It is possible to narrow down and transmit the search packets that arrive at the virtual destination node block earlier in order, and it is possible to prevent useless consumption of resources in the reconfiguration unit. In the reconfiguration unit, a virtual node block through which a predetermined number of search packets have passed can be released, so that the PE constituting the virtual node block is used for route search in the next network or other applications are processed. Can be used in the circuit for

  Further, it is desirable that the units constituting the virtual network further configure each virtual node block to discard the search packet having the same information about the virtual link and / or the virtual node block that has passed through. By discarding search packets that follow a looped route, it is possible to prevent unnecessary search packets from being exchanged between virtual node blocks, thereby increasing processing speed and reducing power consumption. .

Another aspect of the present invention is a method for searching a path from a source node to one or more destination nodes using a processor in a network including a plurality of nodes and a plurality of links connecting the nodes. . The processor may include a reconstruction unit. The reconfiguration unit includes a plurality of processing elements and a connection matrix capable of dynamically changing the connection of the plurality of processing elements, and can dynamically reconfigure a circuit.

The method includes the following steps.
Based on the information of the network collected by 1.1 or more routing protocols, including a plurality of virtual nodes blocks corresponding to the plurality of nodes included in the network, and a plurality of virtual links corresponding to a plurality of links Realizing a virtual network with a processor. In the case of including a reconfiguration unit, a plurality of virtual node blocks respectively corresponding to a plurality of nodes included in the network are configured by a plurality of processing elements, and a plurality of virtual links respectively corresponding to the plurality of links are formed by a connection matrix. Configuring and configuring a virtual network in the reconfiguration unit.
2. Of the plurality of virtual node blocks, a virtual source node block corresponding to the source node transmits a search packet to all virtual links connected to the virtual source node block.
3. Each of the plurality of virtual node blocks adds, to the received search packet, progress information including information on the virtual link and / or virtual node block through which the received search packet has passed, and a search packet in which the received search packet is duplicated To transmit to all virtual links connected to each virtual node block, except for the virtual link where the received search packet has arrived.
4). Of the plurality of virtual node blocks, a route that matches a predetermined condition is selected from the plurality of routes from the source node to the destination node based on the progress information of the plurality of search packets that have arrived at the virtual destination node block corresponding to the destination node. thing.

  The progress information includes bitmap data that can identify a virtual link and / or a virtual node block that has been passed by a bit position, and selecting a path (step 4) is a bit unit logic of the bitmap data of a plurality of progress information. This includes selecting a disjoint path from a plurality of paths from the source node to the destination node by calculating the product.

  The plurality of processing elements include a plurality of delay elements whose delay amounts can be dynamically set, and configuring a virtual network (step 1) includes configuring a plurality of virtual links by the connection matrix and the plurality of delay elements. The delay element included in each virtual link includes setting a delay amount corresponding to the cost of the corresponding link.

  Selecting a route (step 4) matches a predetermined condition from a plurality of routes from the source node to the destination node based on the progress information of the plurality of search packets from the earliest arrival in the virtual destination node block. Including selecting a route.

  Sending to all virtual links connected to each virtual node block (step 3) means that each virtual node block only copies a plurality of search packets from the earliest in order of arrival at each virtual node block. Including transmission to all the virtual links connected to the respective virtual node blocks except for the virtual link on which the received search packet has arrived.

  Sending to all virtual links connected to each virtual node block (step 3) includes discarding search packets with the same information about the virtual links and / or virtual node blocks that have passed.

  FIG. 1 shows an outline of a router having a function of updating a routing table as an example of a system for performing a route search. The router 100 is a device constituting one node of the computer network, and includes appropriate hardware resources including a CPU, a memory, a processor capable of reconfiguring a circuit, and appropriate software resources for controlling them. It has. The router 100 generates and analyzes a network interface 101 capable of exchanging data with an adjacent router through an appropriate transmission path including wired or wireless, and packet data (packet) transmitted / received via the network interface 101. It includes an analysis unit 102 and a routing unit 80 having a function of selecting a next hop router (next hop) to send a packet based on information of the IP header of the packet.

  The functions of the routing unit 80 are provided by dedicated modules or hardware resources shared with other functions, such as a CPU or a reconfigurable processor and software. The routing unit 80 includes neighboring routers by means of a function 82 for selecting a next hop to send a packet with reference to the routing table 81 and an appropriate routing protocol that enables dynamic routing table updating, such as OSPF. And a function 83 for obtaining information on the network, in particular, information indicating the configuration of the network related to routing, and updating the routing table 81. This updating function 83 uses the network information acquired by the routing protocol as a network to be analyzed, and generates a network matrix 89 indicating the configuration of the network.

  In a link state type routing algorithm such as OSPF, link information is periodically exchanged between routers by LSA (Link-State Advertisement). Therefore, each node belonging to the network knows all link information (cost) in the network. Accordingly, the network matrix 89 can include link cost information in the network obtained by LSA. The network matrix 89 can further include network information acquired by another routing protocol together with or in place of OSPF. In this example, the network matrix 89 includes a plurality of nodes included in the target network, a plurality of links connecting the plurality of nodes, and the cost of each link. In a computer network such as the Internet, a typical factor that determines a cost (link cost) between nodes is a reciprocal of a communication speed. In addition, in an optical network, a plurality of factors such as the success rate of wavelength reservation, the type of data to be transmitted, and the bandwidth reserved by QoS can be reflected as the link cost. The network matrix 89 may include elements (routing metrics) that can compare the superiority and inferiority of routes other than the cost.

  FIG. 2A shows an example of the network matrix. This network matrix MN corresponds to the network N shown in FIG. 2B, and the network N includes six nodes A to F. The network matrix MN includes, as elements, the cost of each link having each node included in the network N as a base and another node connected to the base node as a top. The infinite element indicates a node that is included in the network matrix MN and is not connected through a direct route. The value indicating that the node is not connected is not limited to infinity.

  The routing unit 80 includes a system (route search system) 30 that performs route search based on network information (network matrix) 89 collected using a routing protocol. The route search system 30 includes a device (reconfigurable device) 1 having an area (reconfiguration unit) in which a circuit can be dynamically reconfigured. Then, the route search system 30 changes the network configuration corresponding to the network (target network) that is trying to find a route to the virtually reconfigurable device 1 within the range for the purpose of route search based on the network matrix 89. Reconfigure and route search.

  FIG. 3A shows an example of a reconfigurable device. This device 1 is a semiconductor integrated circuit device called DAPDNA developed by the applicant of the present application. The device 1 includes a RISC core module 2 called DAP and a dynamic reconfigurable data flow accelerator 3 called DNA. In addition to DAP2 and DNA3, the device 1 has a DNA4 direct input / output interface 4, a PCI interface 5, an SDRAM interface 6, a DMA controller 7, and other peripheral devices 8, and a high speed for connecting them. Switching bus (internal bus) 9. The DAP2 includes a debug interface 42a, a RISC core 42b, an instruction cache 42c, and a data cache 42d. DNA3 is a PE matrix 10 in which 376 PEs (PEs, processing elements) are two-dimensionally arranged, and the PE matrix 10 is reconfigured by changing the function of PEs included in the PE matrix 10 and / or the connection of PEs. And a configuration memory 19 in which configuration data 18 is stored. The PE 10 includes a plurality of processing elements PE (PEs) and a connection matrix that can dynamically change the connection of the plurality of PEs, and corresponds to a reconfiguration unit that can dynamically reconfigure a circuit. The number of PEs arranged in the PE matrix 10 is not limited to the above.

  The configuration memory 19 has a plurality of banks. For example, as shown in FIG. 3B, in the PE matrix 10, a first function (data flow, circuit design) 17a is configured by configuration data 18 stored in the foreground bank. Further, the second function 17b and the third function 17c are configured by configuration data stored in different background banks, respectively. By switching the bank of the memory 19, the PE matrix 10 is reconfigured with the second function 17b or the third function 17c instead of the first function 17a. The reconstruction of the PE matrix 10 is dynamically performed in one cycle (clock cycle), for example.

  FIG. 3C is an example in which a circuit is reconfigured in the PE matrix 10. A certain application, for example, an MPEG decoder is time-divided into a plurality of functions (subfunctions), and these functions are dynamically reconfigured in a time-division manner into the PE matrix 10, and the functions of the MPEG decoder are assigned to dedicated circuits (dedicated hardware ). Similarly, in the route search, when the number of nodes included in the target network is enormous, the network is divided into a plurality of sub-networks, and the virtual networks corresponding to these sub-networks are dynamically reconfigured in a time division manner. It is possible. Therefore, the route search system 30 employing this device 1 can be applied to a large-scale network. By using such a device, an application requiring a large amount of hardware resources can be executed with a small amount of hardware resources using the device 1 which is a reconfigurable data processing apparatus.

  FIG. 3D shows another example of configuring a circuit in the PE matrix 10. For example, the PE matrix 10 can be reconfigured so that a plurality of functions are realized in order to execute applications having different playback methods. Similarly, in the router 100, the device 1 can be used not only as the route search system 30 but also as a function 82 for selecting a next hop for sending a packet with reference to the routing table 81. Through such use, many applications can be executed using the common hardware (device) 1.

  Further, the device 1 can be implemented by switching many functions at a data flow level (data path level, hardware level) instead of a program level (instruction level). Therefore, processing can be performed at a speed comparable to that of dedicated hardware. Therefore, by reconfiguring the virtual network for route search into the PE matrix 10, it is possible to obtain a processing speed comparable to that configured with dedicated hardware. On the other hand, since the PE matrix 10 can be dynamically reconfigured, it is possible to flexibly follow a change in a network to be searched for a route without delay.

  The PEs included in the PE matrix 10 can perform processing synchronized with a clock, and a data flow (data path) suitable for pipeline processing can be reconfigured in the PE matrix 10. Furthermore, it is also possible to reconfigure a plurality of data flows that perform parallel processing with a large number of PEs. Therefore, by reconfiguring the virtual network in the PE matrix 10 and flowing a route search packet (search packet), a plurality of search packets can be transmitted / received in parallel in the virtual network, similar to the target network. It becomes possible. Furthermore, since the search packet transmission / reception (flow) is processed synchronously, the timing at which the search packet reaches the target node (for example, the destination node) is regarded as information on the quality of the route through which the search packet has passed. It can be used as information for selecting a route. Thus, by dynamically reconfiguring the PE matrix 10 as a virtual network as a circuit, it is possible to provide a route search environment in which high-speed processing speed by hardware and flexibility such as software are fused.

  FIG. 4 shows an enlarged view of the PE matrix 10. The PE matrix 10 connects a plurality of processing elements (PE) 45 arranged two-dimensionally in the vertical and horizontal directions, wirings 47 arranged in a lattice pattern between them, and vertical and horizontal wirings 47 at connection points of the wirings 47. And a switching unit 46 that can be freely switched. The wiring 47 and the switching unit 46 constitute a connection matrix 44 that can freely change the connection of the plurality of PEs 45. The connection matrix 44 may include PEs for adjusting the connection environment, such as a delay PE described below.

  An example of the PE 45 constituting the reconfigurable PE matrix (circuit area) 10 is a function whose function can be freely set by a lookup table or the like. The matrix 10 of the integrated circuit device 1 includes an arithmetic logic operation element, a delay element, a memory element, an element for generating an address for inputting or outputting data, and an element for inputting or outputting data. The elements of the internal configuration suitable for a specific function or processing are arranged. Redundancy is reduced by arranging elements divided into a certain number of functional groups, and AC characteristics and processing speed can be improved.

  Configuration data is supplied from the processor 2 or the configuration memory 19 to each PE 45 via the control bus 9a. With this configuration data, the function of each PE 45 and the connection of the wiring group 47 are controlled, and various data flows (data paths) can be freely configured in the matrix 10. Further, the PE matrix 10 is connected to the external world such as the external memory interface 6, and therefore, an input buffer 48 and an output buffer 49 for inputting / outputting data to be processed, and a bus switching unit (bus interface, BSU) 9b.

  FIG. 5 is an example of PE45. This PE is used for operations having a configuration suitable for arithmetic operations and logical operations. The PE 45 includes an internal data path area 45b whose function can be changed and a control unit 45a for setting the function of the internal data path area 45b. The internal data path unit 45b includes a shift circuit SHIFT and a mask circuit MASK for extracting desired input data from the wiring 47, an input register (FF) for latching the input data dix and diy by a clock signal, and a logical operation A unit ALU and an output register (FF) for latching output data do output to the wiring 47 by a clock signal are provided.

  The control unit 45a of each PE receives configuration data from the processor 2 via the control bus 9a and controls the configuration of the internal data path unit 45b. Therefore, a node having a function of transferring a search packet under a predetermined condition can be reconfigured by one or a plurality of PEs 45 according to configuration data.

  FIG. 6 shows another example of PE45. This PE is suitable for processing for delaying the timing at which data is transmitted. The internal data path unit 45d includes a delay circuit 45e configured by a combination of a plurality of selectors and flip-flops, an input-side flip-flop (FF), an output-side flip-flop (FF), and a selector for selecting a circuit. 45 s. In this example, the delay circuit 45e can set a delay of 0 to 5 clocks according to the configuration data. Therefore, a delay of 1 to 7 clocks can be controlled for each input. Furthermore, it is possible to connect two input systems (X system and Y system) in series according to the configuration information, and to control twice the delay time. In addition, carry signals cix and ciy guided by the row wiring group and the column wiring group for carry signals are output with a delay through a similar data path.

  The PE matrix 10 can be reconfigured so that the delay PE 45 is included in a virtual link connecting virtual nodes of the virtual network. Since the delay amount of the delay PE 45 can be set to be variable by configuration data, the history of the search packet is determined by using the delay amount as a search packet delay (transfer time delay) as a factor that determines link superiority or inferiority. (Progress information).

  As shown in FIG. 1, the route search system 30 includes a configuration unit 31 that forms a virtual network 35 in a PE matrix 10 that is a reconfiguration unit, and a route that matches a predetermined condition from a route searched using the virtual network. And a route selection unit 32 for selecting. The configuration unit 31 reconfigures the virtual network 35 in the PE matrix 10 based on network information (network matrix) 89 collected by one or more routing protocols supported by the router 100. The route selection unit changes from the source node to the destination node based on the progress information of the plurality of search packets that arrived at the virtual destination node block corresponding to the destination node (destination node) among the plurality of virtual node blocks included in the virtual network 35. A route that meets a predetermined condition is selected from a plurality of routes. In this example, the functions as these units 31 and 32 are provided by a program loaded in the DAP 2. These units 31 and 32 may be reconfigured in the PE matrix 10 together with the virtual network 35 or by time division.

  FIG. 7 shows an example of a target network to be a route search target. FIG. 8 shows an example of a virtual network reconfigured in the PE matrix 10 for route search so as to correspond to the network. The target network 21 includes 11 nodes N1 to N11 and 13 links L1 to L13 connecting these nodes. The virtual network 35 includes virtual node blocks PN1 to PN11 corresponding to the nodes N1 to N11, and virtual links PL1 to PL13 respectively corresponding to the links L1 to L13. The virtual node blocks PN1 to PN11 are configured by one or a plurality of PEs 45, and the virtual links PL1 to PL13 are configured by a connection matrix 44 that connects the PEs 45. The virtual links PL1 to PL13 may include not only the wiring 47 and the switching matrix 46 but also a delay element. In the following, when a virtual node block and a virtual link are generally indicated, they are referred to as a virtual node block PN and a virtual link PL.

  The configuration unit 31 generates configuration data 18 for reconfiguring the virtual network 35 in the PE matrix 10 based on the network matrix 89 and supplies the configuration data 18 to the PE matrix 10 or the configuration memory. Thus, the circuit (data flow, data path) functioning as the virtual network 35 is reconfigured in the PE matrix 10. Furthermore, the configuration unit 31 includes all virtual links in which the virtual source node block PN1 corresponding to the source node N1 among the virtual node blocks PN1 to PN11 is connected to the virtual source node block PN1, in this case, the virtual links PL1 and PL2 The configuration data 18 is generated so that the search packet SP is transmitted to the virtual source node block PN.

  In addition, the configuration unit 31 has received each virtual node block PN excluding the virtual source node block N1 and the virtual destination node block N8 among the plurality of virtual node blocks PN1 to PN11 in the received search packet SP. Configuration data 18 is generated so as to add progress information including information of the virtual link and / or virtual node block through which the search packet SP has passed, and each virtual source node block PN is reconfigured. The configuration unit 31 further transmits the search packet SP obtained by duplicating the received search packet SP to all the virtual links PL connected to each virtual node block PN except for the virtual link PL from which the received search packet SP has arrived. Thus, the configuration data 18 is generated, and each virtual source node block PN is reconfigured.

  The configuration unit 31 further sets each virtual node block PN to a copy of a plurality of search packets SP from the earliest in order of arrival at each virtual node block PN, and the virtual link PL from which the received search packet has arrived. Except for this, the configuration data 18 is generated so as to be transmitted to all the virtual links PL connected to each virtual node block PN, and the virtual source node block PN is reconfigured.

  The configuration unit 31 further configures the configuration data 18 to configure each virtual node block PN to discard search packets SP having the same information about the passed virtual link PL and / or virtual node block PN. Generate and reconfigure the virtual source node block PN.

  FIG. 9 shows an example of the search packet SP. This search packet SP includes header information 24 and progress information 25. The progress information 25 includes bitmap data that can identify the virtual link PL and / or the virtual node block PN through which the search packet SP has passed by the bit position. In the following example, the progress information 25 is bitmap data that can identify the virtual link PL through which the search packet SP has passed by the bit position. It is possible to include the bitmap data that can identify the virtual node block PN through which the search packet SP has passed in the progress information 25 by the bit position, similarly to the virtual link. The progress information 25 of the search packet SP that has reached the virtual destination node block PN8 contains information on the virtual link PL through which the search packet SP has passed before reaching the virtual destination node block PN8 from the virtual source node block PN1. . Therefore, the path selection unit 32 can know the path from the virtual source node PN1 to the virtual destination node block PN8 by accessing the progress information 25 of the search packet SP that has reached the virtual destination node block PN8. The route corresponds to a route in the target network 21, and the routing table 81 can be updated by finding a route that meets a predetermined condition.

  One typical route searched is the shortest route from the source node N1 to the destination node N8. For example, by making the delays of the virtual links L1 to L13 constant, the route of the search packet SP that has reached the virtual destination node block PN8 earliest can be selected as the route with the minimum number of hops. In addition, by setting the delay of the virtual links L1 to L13 based on the link cost, the route selection unit 32 can select the route of the search packet SP that has reached the virtual destination node block PN8 earliest as the route with the lowest cost.

  A different example of a typical searched route is a disjoint route from the source node N1 to the destination node N8. The route selection unit 32 calculates a plurality of bitwise ANDs of the bitmap data of the progress information 25 of the plurality of search packets SP that have reached the virtual destination node block PN8, and thereby performs a plurality of operations from the source node N1 to the destination node N8. A disjoint route can be selected from the routes. If the bitwise logical product of the bitmap data of the progress information 25 is all 0, the path included in the compared progress information 25 is a disjoint path. Comparison of bitmaps (bit strings) can be processed in a very short time by using PE for arithmetic operation included in the PE matrix 10. Bitmaps may be compared by software using DAP2.

  Further, by making the delay of the virtual links L1 to L13 constant, the route selection unit 32 can select a disjoint route with the minimum number of hops at the timing when the virtual destination node block PN8 is reached. By setting the delays of the virtual links L1 to L13 based on the link cost, the route selection unit 32 can select a disjoint route with the minimum cost based on the timing of the search packet SP that has reached the virtual destination node block PN8.

  FIG. 10 shows an example of a method for searching for a route from a source node to one or a plurality of destination nodes using the PE matrix 10 which is a reconfiguration unit. In step 71, network information 89 is collected by one or more routing protocols provided by the router 100. In step 72, the configuration unit 31 configures the virtual network 35 in the PE matrix 10 based on the network information 89. In step 73, the search packet SP is transmitted from the virtual source node block PN1 corresponding to the source node N1 to all the virtual links PL1 and PL2 connected to the virtual source node block PN1. In the progress information 25 at this stage, all bits are 0.

  In step 74, when the virtual node block PN receives the search packet SP, the virtual node block PN adds it to the progress information 25 including information on the virtual link through which the received search packet SP has passed. For example, the virtual node block PN2 sets the first bit of the bitmap of the progress information 25 to 1, thereby adding to the progress information 25 that the virtual link PL1 has been passed. Then, the search packet SP including the progress information 25 is duplicated and transmitted to all the virtual links PL connected to the virtual node block PN except for the virtual link PL1 from which the received search packet SP has arrived. For example, the virtual node block PN2 generates two copies of the search packet SP in which the first bit of the bitmap of the progress information 25 is set to 1, and transmits the duplicates to the virtual links PL3 and PL4.

  In step 75, the search packets SP are transferred between the virtual node blocks PN until a sufficient number of search packets SP reach the virtual destination node block PN8. When a sufficient number of search packets SP reach the virtual destination node block PN8, in step 76, the route selection unit 32 reaches from the source node N1 to the destination node N8 based on the obtained progress information 25 of the plurality of search packets SP. Multiple routes can be obtained. Then, a route that matches a predetermined condition is selected from the obtained routes.

  In step 75, when a disjoint path is selected, as described above, it is desirable to calculate the logical product of bitmap data of a plurality of pieces of progress information 25. The search for the shortest route with the minimum number of hops and the shortest route with the minimum cost can be performed together with the search for the disjoint route. It is possible to select the shortest route in a disjoint route based on the progress information 25 of only a plurality of search packets SP (arriving first) (from the earliest) to the virtual destination node block PN8. is there. Basically, the purpose of route search is to select the shortest route in consideration of various conditions, and there is no need to wait for all search packets SP to reach the virtual destination node block PN8.

  In step 74, each virtual node block PN preferably transfers only a plurality of copies of the search packet SP from the earliest arrival order. The purpose of route search is to select the shortest route in consideration of various conditions, and it is not usually necessary to make use of the search packet SP that follows the link in vain. Further, in step 74, it is desirable to discard the search packet SP having the same information about the passed virtual link (or virtual node block) in the progress information 25. For example, when a search packet SP having information that has passed through the virtual link PL3 reaches the virtual node block PN4 through the virtual link PL3, it means that the search packet SP is looping through the virtual link PL3. The search packet SP indicating the route to be looped is not necessary for the route search. It is desirable to discard such a search packet SP and reduce the number of search packets SP flowing through the virtual network 35.

  FIGS. 11 to 17 show how the route is searched for by the route search system 30 in the network 21 shown in FIG. First, link numbers (L1 to L13) are assigned to all links in the network 21. The link number may be similar to an SRLG (Shared Risk Link Group) number, but in this example, the link number promotes independence.

  In step 1 shown in FIG. 11, a search packet SP is sent to all nodes connected from the S (source) node N1 (PN1) to the source node N1, and the search is started.

  In Step 2 shown in FIG. 12, the node N (PN) where the search packet SP has arrived is connected to the node N, and the search is performed by sending the search packet SP to all the nodes N that do not return backward. At this time, each search packet SP holds the progress information 25 including the passed link number.

  As shown in FIGS. 13 to 16, in steps 3 to 6, similarly to the above, a search is performed for the next node N (PN) every cycle. In step 5 shown in FIG. 15, the search packet SP that created the loop is discarded. As a loop detection method, a method of judging from the passing node number of the progress information 25, an appropriate identification information is added to the search packet SP, and the search packet SP that each virtual node block PN has already passed is grasped. A method of discarding when searching for the virtual link PL again can be considered. It is possible to set the delay of each virtual link PL to be the same. In this case, in the virtual network 35 reconfigured in the PE matrix 10, if the latency in the virtual node block PN is different, the following is performed every cycle. The virtual node block PN can be searched. Therefore, the route search of the network 21 including a large number of nodes and links can be performed in a short time.

  Even when the cost is set for each virtual link PL by the delay element, a delay of only a few clocks occurs for one virtual link PL. Compared with the case where a packet for route search is transmitted and received after waiting for the network protocol to process, the route search of the network 21 can be performed in a much shorter time. Further, since the virtual network 35 exists inside the route search system 30, a route is searched based on the progress information 25 without returning the search packet SP that has reached the virtual destination node block PN8 to the source node block PN1. it can. In this respect as well, a route search can be performed in a much shorter time than when a packet is sent to the target network 21.

In step 7 shown in FIG. 17, the search is completed for all routes. An example of the route information 25 included in the search packet SP that has reached the virtual destination node block PN8 is as follows in the order of arrival. In parentheses, link numbers are shown without L.
Route 1 (1,4,7) 3hop
Route 2 (2, 5, 12, 7) 4hop
Route 3 (1, 3, 6, 13) 4hop
Route 4 (2, 5, 8, 9, 10, 11) 6hop
Route 5 (1, 4, 12, 8, 9, 10, 11) 7hop
Route 6 (2, 5, 12, 4, 3, 6, 13) 7hop

  Therefore, in the route selection unit 32, it can be seen that route 1 is the shortest route with the minimum number of hops. Further, the route selection unit 32 can obtain a combination of disjoint routes among these six routes.

  FIG. 18 shows the progress information 25 included in the search packet SP that has reached the virtual destination node block PN8 by following each route. The progress information 25 is bitmap data having a bit width of at least 13 bits, and the bit at the position corresponding to the virtual link PL that has passed is 1. For example, if the network includes about 100 links, the progress information 25 needs to be 100 bits wide and is about 13 bytes of data.

  When a disjoint route is obtained with respect to route 1, the logical product (AND) of bit unit of the progress information 25 of route 1 and the progress information 25 of all other routes (routes 2 to 6) is calculated. It ’s fine. If even one logical product is “1”, it is not disjoint.

  FIG. 19 is a flowchart showing one method for obtaining a logical product by software. First, the link number j (1-13) used by the route i (1-6) is assumed to be an l (i, j) matrix. Next, one route (for example, route 1) is selected, and logical product (AND) is taken for each link number of all other routes (for example, routes 2 to 6), and even if one logical product is “1”. For example, it is determined that it is not a disjoint.

As a result, the route selection unit 32 can obtain the following disjoint combinations of routes.
Route 1 and Route 4
Route 2 and Route 3
Route 3, route 2 and route 4
Route 4 and Route 3
Routes 5 and 6 do not have disjoint routes.

Furthermore, the route selection unit 32 can calculate the following three disjoint route candidates.
Route 1 (3 hop) + Route 4 (6 hop) = 9 hop
Route 2 (4 hop) + Route 3 (4 hop) = 8 hop
Route 3 (4 hop) + Route 4 (6 hop) = 10 hop
In this case, it can be seen that the combination of route 2 and route 3 has the smallest number of hops. Therefore, in this case, the route selection unit 32 selects the combination of the route 2 and the route 3 as an ideal solution. These routes 2 and 3 are link disjoint as shown in FIG. 20, and are also node disjoint routes. By selecting a node disjoint route, it is possible to provide a routing table 81 with high reliability not only for links but also for node defects. When the node disjoint route is positively selected, it is desirable to include the identification information of the passed node (virtual node block) in the progress information 25. In addition, the condition for selecting a disjoint route is not limited to the total number of hops, but may be a combination that minimizes the total number of costs if the cost is reflected. Alternatively, a combination that minimizes the cost difference or hop difference of disjoint routes may be used.

  In a large-scale network, waiting for searching for all routes has a problem that the search time becomes long. Accordingly, the route selection unit 32 selects and selects only the best m routes of the search packet SP that arrives at the virtual destination node block PN8 in the early order.

  In addition, in order to obtain the best m in the order of arrival in the virtual destination node block PN8, each relay node block PN processes only m pieces in the order of arrival and discards the rest. Thus, the upper limit of the search in the network is determined by the maximum m and n (n is the number of nodes).

  Further, when hardware resources such as PEs in the PE matrix 10 are insufficient, the virtual network 35 can be multi-contexted and reconfigured in the PE matrix 10 in a time division manner. Alternatively, the PE matrix 10 of a plurality of devices 1 may be directly connected to virtually expand the hardware resources of the PE matrix 10.

  In the route search system 30, if a virtual network can be reconfigured in the PE matrix 10 as a reconfiguration unit, route search is possible. Therefore, the present invention can also be applied to route search in a range that covers a plurality of networks or a plurality of ASs to which different routing protocols are applied.

The figure which shows schematic structure of a router. FIG. 2A shows an example of a network matrix, and FIG. 2B shows an example of a corresponding network. 3A shows a schematic configuration of an example of a reconfigurable device, FIG. 3B shows an overview of the PE matrix, and FIGS. 3C and 3D show the PE matrix. The figure which shows a mode that it reconfigures dynamically. The figure which shows the structure of PE matrix. The figure which shows schematic structure of an example of PE. The figure which shows schematic structure of the other example of PE. The figure which shows an example of the network of the target of route search. The figure which shows an example of a virtual network. The figure which shows an example of a search packet. The flowchart which shows the process of route search. The figure which shows an example (step 1) of route search. The figure which shows step 2 of route search. The figure which shows step 3 of route search. The figure which shows step 4 of route search. The figure which shows step 5 of route search. The figure which shows step 6 of a route search. The figure which shows step 7 of a route search. The figure which shows the bitmap of progress information. The flowchart which shows an example of the process which calculates | requires a logical product. The figure which shows an example of the combination of a disjoint route.

Explanation of symbols

1 reconfigurable device, 10 PE matrix (reconfigurable unit)
30 Route Search System 31 Virtual Network Configuration Unit, 32 Route Selection Unit 35 Virtual Network 80 Routing Unit 100 Router

Claims (11)

A system for searching for a route from a source node to one or more destination nodes in a network including a plurality of nodes and a plurality of links connecting them,
A plurality of virtual node blocks respectively corresponding to the plurality of nodes included in the network , and a plurality of virtual links respectively corresponding to the plurality of links, based on the network information collected by one or a plurality of routing protocols ; Including a virtual network configuration unit that realizes a virtual network including:
The virtual network configuration unit configures a virtual source node block corresponding to the source node among the plurality of virtual node blocks so as to transmit a search packet to all virtual links connected to the virtual source node block, further,
Each of the plurality of virtual node blocks is added to the received search packet with progress information including information on the virtual link and / or virtual node block through which the received search packet has passed, and the received search packet is duplicated. Configured to transmit the search packet to all the virtual links connected to the respective virtual node blocks, excluding the virtual link from which the received search packet has arrived,
Furthermore, the system
Among the plurality of virtual node blocks, the progress information of the plurality of search packets that have arrived at the virtual destination node block corresponding to the destination node is set to a predetermined condition from a plurality of routes from the source node to the destination node. A system having a route selection unit for selecting a matching route. The progress information according to claim 1, wherein the progress information includes bitmap data that can identify the passed virtual link and / or virtual node block by a bit position,
The path selection unit selects a disjoint path from a plurality of paths from the source node to the destination node by calculating a bitwise logical product of the bitmap data of the plurality of pieces of progress information. system. 3. The reconfiguration unit according to claim 1, wherein the processor is a reconfiguration unit capable of dynamically reconfiguring a circuit, and a plurality of processing elements and a connection matrix capable of dynamically changing the connection of the plurality of processing elements. Having a reconstruction unit including
The system in which the virtual network configuration unit configures the plurality of virtual node blocks by the plurality of processing elements and configures the plurality of virtual links by the connection matrix. In Claim 3 , the plurality of processing elements include a plurality of delay elements that can dynamically set the delay amount,
The units constituting the virtual network configure the plurality of virtual links by the connection matrix and the plurality of delay elements, and the delay elements included in the respective virtual links correspond to the cost of the corresponding links. Set the amount of delay to be performed. 5. The system according to claim 3, wherein the reconfiguration unit can reconfigure the circuit including the plurality of processing elements and the connection matrix on a cycle basis. The system according to any one of claims 3 to 5, wherein the virtual network configuration unit configures the virtual network into the reconfiguration unit in a time-sharing manner by making the virtual network into multiple contexts. More in any one of claims 1 to 6, wherein the path selection unit, by the progress information of a plurality of said search packet from those early in the order they arrived to the virtual destination node block, extending from the source node to the destination node A system that selects a route satisfying a predetermined condition from the routes of the system. In any one of Claims 1 thru | or 7 , the unit which comprises the said virtual network makes only each copy of a several search packet from the thing with the early order which arrived at each said virtual node block to the said each virtual node block, A system configured to transmit to all virtual links connected to the respective virtual node blocks except for the virtual link on which the received search packet has arrived. In any one of claims 1 to 8, units constituting the virtual network, the respective virtual node block, so as to discard the search packet having the same information for the virtual link and / or virtual nodes block and the passage Configure the system. A method for searching a path from a source node to one or more destination nodes using a processor in a network including a plurality of nodes and a plurality of links connecting the nodes,
A plurality of virtual node blocks respectively corresponding to the plurality of nodes included in the network , and a plurality of virtual links respectively corresponding to the plurality of links, based on the network information collected by one or a plurality of routing protocols ; All virtual nodes in which a virtual source node block corresponding to the source node among the plurality of virtual node blocks of the virtual network realized by the processor is connected to the virtual source node block. Sending a search packet on the link;
Each of the plurality of virtual node blocks adds, to the received search packet, progress information including information on the virtual link and / or virtual node block through which the received search packet has passed, and replicates the received search packet. Sending the search packet to all virtual links connected to the respective virtual node blocks, excluding the virtual link from which the received search packet has arrived;
Among the plurality of virtual node blocks, the progress information of the plurality of search packets that have arrived at the virtual destination node block corresponding to the destination node is set to a predetermined condition from a plurality of routes from the source node to the destination node. Selecting a matching route. The processor according to claim 10, wherein the processor includes a reconfiguration unit including a plurality of processing elements and a connection matrix capable of dynamically changing a connection of the plurality of processing elements, and dynamically reconfiguring a circuit. ,
Before the transmission, the method
Configuring the plurality of virtual node blocks by the plurality of processing elements, configuring the plurality of virtual links by the connection matrix, and configuring the virtual network in the reconfiguration unit.

Images