JP2006270677A - Optimum path arrangement searching apparatus and optimum path arrangement searching method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an apparatus and a method for searching an optimum optical path structure by using genetic algorithm in changing logic topology built on a network in accordance with traffics. <P>SOLUTION: First, a plurality of gene codes each indicating a node arrangement on the logic topology are generated. Each of traffics indicated by each of the gene codes of the logic topology is evaluated. A selectable probability that the lower a result of evaluation is, the smaller the traffic is, is assigned to each of the gene codes. Selection according to the selectable probability is repeated predetermined times, thereby selecting the predetermined number of gene codes. An optional gene code pair is generated from the selected gene codes, and randomly selected gene constituents are crossed between the gene code pairs and one part or all of the gene constituents contained in each of the gene codes are mutagenized at a predetermined probability. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a technique for searching for an optimal arrangement of paths constituting a logical topology based on communication traffic. The present invention is used in, for example, an optical communication network having a function of dynamically changing a logical topology constructed using WDM (Wavelength Division Multiplex) technology.

  In recent years, in communication networks such as the Internet, an increase in traffic due to a rapid increase in users has become a problem. As a technique for dealing with this problem, WDM has attracted attention. WDM is a technique for multiplexing carrier waves having different optical wavelengths. By using WDM, a plurality of carrier waves can be propagated in parallel on one optical transmission line, so that a large-capacity communication network can be constructed at low cost.

  In an optical communication network using WDM, for example, a node having an optical add / drop multiplexer (OADM) or an optical cross connect (OXC) is used as each node. The optical add / drop function is a function of inserting an optical carrier wave of another wavelength into an optical carrier wave multiplexed using WDM, or branching an optical carrier wave of a specific wavelength from the multiplexed optical carrier wave. The optical cross-connect function is a function for switching a communication path of an optical carrier wave multiplexed using WDM in units of wavelengths. By using these functions, it is possible to construct a logical topology composed of a plurality of optical paths on an optical communication network (that is, a physical topology) constructed on an optical transmission line.

  Further, as an optical communication network using WDM, one having a function of dynamically changing a logical topology is known. In a network, congestion or a failure may occur in some nodes and paths. In such a case, by dynamically changing the logical topology, it becomes possible to cut through the node where congestion has occurred in the communication packet and to bypass the failure occurrence section.

  Here, the cut-through means that a communication packet is passed as it is by using only the optical add / drop function or the optical cross connect function without using the routing function of the node (paragraph 0002 of Patent Document 1 below). ~ 0005 and Figure 6). By setting a cut-through path for a node where congestion has occurred, the processing load on the node can be reduced.

As a method for dynamically changing an optical path to bypass a faulty section, a dedicated protection method and a shared protection method are known. The exclusive protection method is a method in which one optical path is substituted with another optical path. The shared protection method is a method of substituting a plurality of optical paths with another optical path.
JP 2002-374291 A

  As described above, when traffic is concentrated on a specific node, the processing load on the node can be reduced by newly setting an optical path that cuts through the node. However, such a path setting may increase the processing load of other nodes for the following reasons.

  When the setting of a new optical path is changed, routing information of an IP (Internet protocol) layer is changed in the optical communication network. Accordingly, when a new cut-through path is set, if it is often determined that the cut-through path is the optimum route, traffic may concentrate on the cut-through path. In such a case, the processing load of the cut-through node is rarely excessive, but the processing load increases at the node that terminates the cut-through path.

  For this reason, when a new cut-through path is set, it is desired that traffic concentration does not occur in the entire optical communication network in consideration of a route change in the IP layer. However, the number of cut-through path setting patterns that terminate at two specific nodes increases exponentially as the number of wavelengths and the number of nodes used in the optical communication network increase. For this reason, if an attempt is made to select an optimal cut-through path in anticipation of the traffic occurrence state in the entire optical communication network for all setting patterns, the processing load becomes enormous and is not realistic.

  An object of the present invention is to provide an optimum path arrangement retrieval apparatus that can retrieve an optimum optical path structure by a simple process.

  (1) The first invention relates to an optimum path arrangement search apparatus that searches for a path arrangement with a small amount of traffic using a plurality of gene codes indicating a node arrangement on a logical topology.

  Then, an evaluation means for evaluating the traffic of the logical topology indicated by each gene code, and a selection probability corresponding to the selection probability by assigning a selection probability such that the smaller the evaluation result is, the higher the selection result is predetermined. A selection means for selecting a predetermined number of gene codes by repeating a number of times, a crossover means for generating an arbitrary gene code pair from the gene code and exchanging randomly selected gene elements between the gene code pairs, and a gene Mutation means for changing a part or all of the genetic elements included in the code with a predetermined probability.

  (2) The second invention relates to an optimum path arrangement search method for searching a path arrangement with a small traffic using a plurality of gene codes indicating a node arrangement on a logical topology.

  Then, an evaluation step for evaluating the traffic of the logical topology indicated by each gene code, and a selection probability that becomes higher as the evaluation result in the evaluation step is smaller is assigned to each gene code according to the selection probability. A selection step of selecting a predetermined number of gene codes by repeating the selection a predetermined number of times, and generating an arbitrary gene code pair from the gene code selected in the selection step, and randomly selecting gene elements as gene code pairs A crossover step for exchanging between them, and a mutation step for changing a part or all of the genetic elements included in the gene code crossed in the crossover step with a predetermined probability.

  According to the present invention, since a genetic algorithm can be applied to search for a path structure with small traffic, an optimal path structure or a path structure close to it can be searched with a simple process.

  Embodiments of the present invention will be described below with reference to the drawings. In the drawings, the size, shape, and arrangement relationship of each component are shown only schematically to the extent that the present invention can be understood, and the numerical conditions described below are merely examples. .

First Embodiment Hereinafter, a first embodiment of the present invention will be described with reference to FIGS.

  FIG. 1 is a conceptual diagram showing a data plane configuration of an IP backbone network according to this embodiment. As shown in FIG. 1, the data plane of the IP backbone network 100 includes nodes N1 to N9 and optical fibers F1 to F24.

  Each of the nodes N1 to N9 includes an IP router and an optical core node (not shown) (described later).

  The optical fibers F1 to F24 respectively connect and connect physically adjacent nodes. “1” to “24” are assigned to the optical fibers F1 to F24 as fiber IDs for identification.

  With such a configuration, an optical path (Lambda Switch Capable-Label Switched Path; LSC-LSP) having an optical core node as a relay node and an optical wavelength as an identifier can be constructed on the IP backbone network 100 (described later). .

  FIG. 2 is a conceptual diagram showing a control plane configuration of the IP backbone network 100. As shown in FIG. 2, the control plane of the IP backbone network 100 includes a traffic measurement management device 210, an optimum optical path arrangement search device 220, an optical path setting management device 230, and control networks 240 and 250. Yes.

  The traffic measurement management device 210 sends traffic information (inflow packet traffic, outflow packet traffic, etc.) measured by the traffic measurement devices 361 and 362 (see FIG. 3 described later) of the nodes N1 to N9 to the control network 240. To collect and manage.

  The optimum optical path arrangement search device 220 uses the traffic information received from the traffic measurement management apparatus 210 to calculate an optimum optical path arrangement based on a genetic algorithm (described later).

  The optical path setting management device 230 uses the IP router 351 and the optical switch control device 371 (see FIG. 3 described later) of each of the nodes N1 to N9 via the control network 250 based on the calculation result of the optimum optical path arrangement search device 220. By controlling, the optimum optical path is set in the data plane (see FIG. 1).

  FIG. 3 is a block diagram showing an internal configuration of the node N1. FIG. 3 shows an example in which four types of carriers (wavelengths λ1 to λ4) are WDM multiplexed and propagated in the optical path. As shown in FIG. 3, the node N1 includes optical amplifiers 311 to 314, optical multiplexers / demultiplexers 321 to 324, optical switches 331 and 332, optical transceivers 341 to 344, an IP router 351, traffic measuring devices 361 and 362, and An optical switch control device 371 is provided.

  The optical amplifiers 311 to 314 amplify the WDM multiplexed optical carrier wave. As the optical amplifiers 311 to 314, for example, EDFA (Erbium Doped Fiber Amplifier) is used.

  The optical multiplexers / demultiplexers 321 and 322 receive the WDM multiplexed optical carrier wave and separate it into carrier waves having wavelengths λ1 to λ4. On the other hand, the optical multiplexers / demultiplexers 323 and 324 receive the optical carrier waves having the wavelengths λ1 to λ4 and multiplex them (that is, WDM multiplexing). As the optical multiplexer / demultiplexers 321-324, for example, an AWG (Arrayed Waveguide Grating) wavelength multiplexer / demultiplexer is used.

  The optical switches 331 and 332 set an optical carrier path for each of the optical wavelengths λ1 to λ4 between the optical multiplexers / demultiplexers 321 to 324 and the optical transceivers 341 to 344. For example, the optical carrier wave output from the λ1 terminal of the optical multiplexer / demultiplexer 321 is either the λ1 terminal of the optical multiplexer / demultiplexer 323, the λ1 terminal of the optical multiplexer / demultiplexer 324, or the optical transceiver 341 by the optical switches 331 and 332. Sent to. Similarly, the optical carrier wave output from the λ1 terminal of the optical multiplexer / demultiplexer 322 is also sent to either the λ1 terminal of the optical multiplexer / demultiplexers 323, 324 or the optical transceiver 341. On the other hand, the optical carrier wave output from the optical transceiver 341 is sent to one of the λ1 terminals of the optical multiplexer / demultiplexers 323 and 324. The optical carriers of other wavelengths λ2 to λ4 are the same as the carriers of wavelength λ1, and the paths from the optical multiplexer / demultiplexers 321 and 322 to the optical multiplexer / demultiplexers 323 and 324 or the optical transceivers 342 to 344, and the optical transceivers Paths from 342 to 344 to the optical multiplexers / demultiplexers 323 and 324 are set.

  The optical transceivers 341 to 344 perform wavelength conversion of an optical carrier wave transmitted and received between the optical switches 331 and 332 and the IP router 351. That is, the optical transceiver 341 converts the wavelength λ1 of the optical carrier wave input from the optical switches 331 and 332 into the wavelength λ0 used by the IP router 351, and converts the wavelength λ0 of the optical carrier wave input from the IP router 351 into the wavelength λ0. , Λ1. Similarly, the optical transmitter / receiver 342 performs bidirectional wavelength conversion between wavelengths λ2 and λ0, the optical transmitter / receiver 343 performs wavelength conversion between wavelengths λ3 and λ0, and the optical transmitter / receiver 344 performs wavelength conversion between wavelengths λ4 and λ0. When the IP router 351 performs routing or the like using an electrical signal, a photoelectric conversion device may be used as the optical transceivers 341 to 344.

  The IP router 351 is communicatively connected to the optical switches 331 and 332 via the optical transceivers 341 to 344 and is connected to the external networks 381 and 382 via the traffic measuring devices 361 and 362. As described above, communication between the IP router 351 and the optical transceivers 341 to 344 is performed using, for example, an optical carrier wave having a wavelength λ0. On the other hand, the communication between the IP router 351 and the external networks 381 and 382 may use an optical carrier wave having a wavelength λ0, an optical carrier wave having a wavelength λ1 to λ4 or other wavelengths, or an electric signal. May be used. When an optical carrier wave or an electrical signal other than the wavelength λ 0 is used, a wavelength conversion function or a photoelectric conversion function is provided in the IP router 351. The IP router 351 reads an IP packet from the received optical carrier or the like, and performs IP layer routing for the IP packet. That is, the IP router 351 reads the destination address of the IP packet received from the optical transceivers 341 to 344 and the external networks 381 and 382, searches the IP path corresponding to the destination address based on the routing table, and The IP packet is transmitted to the optical transceivers 341 to 344 or the external networks 381 and 382 corresponding to the path. The IP router 351 rewrites an internal routing table (not shown) based on a control signal received from the optical path setting management device 230 (see FIG. 2) via the control network 250.

  The traffic measurement devices 361 and 362 measure the traffic of IP packets transmitted and received between the IP router 351 and the external networks 381 and 382. The measurement result is sent to the traffic measurement management device 210 via the control network 240.

  The optical switch control device 371 sets and changes the optical carrier path provided in the optical switches 331 and 332 based on the control signal received from the optical path setting management device 230 via the control network 250.

  Since the internal configurations of the other nodes N2 to N24 are the same as the internal configuration of the node N1, description thereof will be omitted.

  FIG. 4 is a block diagram showing an internal configuration of the optimum optical path arrangement search device 220 (see FIG. 2). As shown in FIG. 4, the optimum optical path arrangement search device 220 includes a reception processing unit 410, a gene code table 420, an objective function table 430, a random number generation unit 440, a path search unit 450, a physical path allocation. A processing unit 460, a transmission processing unit 470, and a bus 480 are provided.

  The reception processing unit 410 receives traffic information from the traffic measurement management device 210 (see FIG. 2) and sends it to the objective function table 430.

  The gene code table 420 stores gene codes and the like for performing a path search using a genetic algorithm.

  The objective function table 430 stores objective functions for evaluating candidate optical path topologies. In this embodiment, the above-described traffic information and the like are stored in the objective function table 430 as the objective function.

  The random number generation unit 440 generates a random number under a predetermined condition based on the control of the path search unit 450 and the like.

  The path search unit 450 searches for an optimal optical path topology using the above-described traffic information, gene code table, objective function, random number, and the like.

  The physical path assignment processing unit 460 assigns a physical path for actually reflecting the optimum optical path topology obtained by the search by the path search unit 450 on the IP backbone network 100.

  The transmission processing unit 470 sends the result of the allocation processing performed by the physical path allocation processing unit 460 to the optical path setting management device 230 (see FIG. 2).

  The bus 480 connects the components 410 to 470 to each other.

  Next, the operation of the IP backbone network 100 shown in FIGS. 1 to 4 will be described.

  First, the operation of the nodes N1 to N9 will be described by taking the node N1 as an example. Here, a case will be described in which an optical path from the node N2 to the node N1 is set via the optical fiber F1, and an optical path from the node N4 to the node N1 is set via the optical fiber F6.

  As described above, four types of carrier waves (wavelengths λ1 to λ4) are WDM multiplexed and propagated in the optical fiber F1. As a result, the IP packet is transferred from the node N2 to the node N1. The WDM multiplexed light is amplified by the optical amplifier 311, demultiplexed by the optical multiplexer / demultiplexer 321, and sent to the optical switch 331. For example, when an optical path of wavelength λ1 is set between the nodes N2 and N1, the optical carrier wave (wavelength λ1) input to the optical switch 331 is sent to the optical transceiver 341. As described above, the optical transmitter / receiver 341 converts the wavelength of this optical carrier wave to λ0 and sends it to the IP router 351. The IP router 351 reads an IP packet from this optical carrier, and further reads a destination address from the IP packet. Then, the IP router 351 searches for an optimal optical path corresponding to the destination address using an internal routing table (not shown).

  When the searched optical path is an optical path using one of the optical fibers F6 (any one of optical carriers having wavelengths λ1 to λ4), the IP router 351 corresponds to any one of the optical transceivers 341 to 344. The optical carrier wave related to the IP packet is output. The optical transceivers 341 to 344 convert the wavelength λ0 of the optical carrier wave into a corresponding wavelength (any one of λ1 to λ4) and output it. The returned optical carrier wave is sent to the optical multiplexer / demultiplexer 323 via the switch 331 and multiplexed in WDM. The multiplexed optical carrier wave is amplified by the optical amplifier 313 and then transferred to the node N4 via the optical fiber F6.

  If the destination address of the IP packet is an address other than the IP backbone network 100, the optimum optical path is one of the external networks 381 and 382. In this case, the IP router 351 transmits the optical carrier wave related to the IP packet to the corresponding external network.

  On the other hand, when an IP packet is transferred from the external network 381, 382 to the IP backbone network 100, the IP router 351 searches for the optimum optical path from the destination address of the IP packet. Then, the IP router 351 outputs the optical carrier wave related to the IP packet to the optical transceivers 341 to 344 corresponding to the optimum optical path. Further, the optical carrier wave is wavelength-converted by the optical transceiver, sent to one of the optical multiplexers / demultiplexers 323 and 324 via the switches 331 and 332, and after WDM multiplexing and optical amplification, Output to the corresponding optical fiber.

  Traffic between the IP router 351 and the external networks 381, 382 is measured by the traffic measuring devices 361, 362. This measurement result is sent to the traffic measurement management device 210 via the control network 240.

  Next, a case where a cut-through path from the node N2 to the node N4 is set with the node N1 as a relay point will be described.

  As in the case described above, the optical carrier wave received from the node N2 via the optical fiber F1 is amplified by the optical amplifier 311, demultiplexed by the optical multiplexer / demultiplexer 321, and then input to the optical switch 331. When the relay point of the cut-through path of the wavelength is set in the node N1, the optical switch 331 sends the input optical carrier wave to the optical multiplexer / demultiplexer 323 as it is without sending it to the optical transceivers 341-344. Therefore, the optical carrier wave transmitted from the node N2 passes through the node N1 as it is without being subjected to IP routing processing and is sent to the node N4. That is, when such a cut-through path is set, a path is set between the nodes N2 and N1 and between the nodes N1 and N4 in the physical topology, but a path between the nodes N2 and N4 is set in the logical topology. It will only be set.

  FIG. 5 is a conceptual diagram showing a logical topology according to this embodiment. As shown in FIG. 5, by using a cut-through path, a logical path for communication connection between non-adjacent logical arrangements L1 to L9 can be set. Therefore, a network having a data plane configuration as shown in FIG. In addition, a logical topology as shown in FIG. 5 can be constructed. A structure in which both-end logical arrangements in the same row and the same column are connected by a logical path as shown in FIG. 5 is referred to as a torus structure. In general, a logical topology having a torus structure is considered to have a high load balancing effect.

  Arbitrary nodes can be arranged at the positions of the logical arrangements L1 to L9 of this torus structure by appropriately setting a normal optical path and a cut-through path. FIG. 6 is a conceptual diagram showing an example. And the node distribute | arranged to each logical arrangement | positioning L1-L9 can be arbitrarily changed by changing the setting of an optical path or a cut-through path suitably. That is, the change of the logical topology by changing the setting of the optical path or cut-through path can be regarded as the node rearrangement in the torus structure.

  By changing the cut-through path while maintaining the torus structure, it is possible to reduce the amount of relay traffic of the IP router. That is, effective load distribution can be achieved by appropriately rearranging the nodes in the torus structure network. In this embodiment, such node rearrangement is realized using a genetic algorithm.

  Hereinafter, the node rearrangement operation of the IP backbone network 100 according to this embodiment will be described.

  As described above, in each of the nodes N1 to N9, the traffic amount is measured by the traffic measuring devices 361 and 362 (see FIG. 3). The measurement result is sent to the traffic measurement management device 210 (see FIG. 2) via the control network 240. The traffic measurement management device 210 converts the traffic distribution of each of the nodes N1 to N9 into a matrix and sends it to the optimum optical path arrangement search device 220.

  The optimum optical path arrangement search device 220 receives this traffic matrix by the reception processing unit 410 (see FIG. 4). Then, the optimum light path arrangement search device 220 searches for the optimum light path arrangement using a genetic algorithm. Here, the genetic algorithm is a kind of arithmetic algorithm for obtaining an optimal solution for a certain problem, and was devised by imitating the genetic laws of the living world. In the genetic algorithm, it tries to obtain the desired “evaluation” by repeating the operations of “selection”, “crossover”, and “mutation”, which are arithmetic processes imitating the laws of genetics, to evolve the solid group. . Here, the individual group is an individual, that is, a set of operation solutions. By using a genetic algorithm, an optimal solution or a solution close thereto can be searched globally.

  Hereinafter, the calculation operation of the optimum optical path arrangement search device 220 will be described with reference to FIGS.

  FIG. 7 is a flowchart schematically showing the arithmetic processing of the optimum optical path arrangement search device 220.

(1) Generation of initial population First, an initial population consisting of a plurality of gene codes is generated (see step S701 in FIG. 7). The gene code is data that is a candidate for an arithmetic solution. In this embodiment, a genetic algorithm is used to search for the optimal rearrangement of nodes in the logical topology of the torus structure. As described above, the change of the logical topology is actually realized by changing the setting of the optical path or cut-through path. However, in the search process of the optimum optical path arrangement search device 220, it is considered that the node is rearranged (see FIG. 5). The IP backbone network 100 of this embodiment includes nine nodes N1 to N9 (see FIG. 1). Thus, the genetic code comprises 9 elements.

  FIG. 8 is a table showing the data structure of the gene code used in this embodiment. As shown in FIG. 8, the gene code used in this embodiment has nine node IDs listed in the order of logical arrangement. Here, the logical arrangement is an arrangement position of each node in the reference logical topology (see FIGS. 5 and 6). The node ID is a number assigned to each of the nodes N1 to N9. For example, the ID of the node N1 is “1” and the ID of the node N2 is “2”. Note that the gene code shown in FIG. 8 matches the node arrangement of the logical topology shown in FIG.

  In this embodiment, the case where the initial population includes 20 gene codes will be described as an example. However, the number of gene codes included in the initial population is arbitrary and may be determined according to the reliability required for the calculation result. The value of each gene code included in the initial population is also arbitrary. In this embodiment, these gene codes are generated by the random number generation unit 440 (see FIG. 4). The generated gene code is sent to the gene code table 420 via the bus 480. The gene code table 420 stores 20 gene codes (that is, 20 kinds of node arrangement patterns) as an initial population.

(2) Evaluation of gene code Next, the path search unit 450 evaluates the gene code. Evaluation is processing for calculating a parameter for determining whether or not a gene code satisfies a predetermined condition. As will be described below, in the evaluation process of this embodiment, the relay traffic of the logical topology indicated by each optimum gene code is evaluated (see steps S702 and S703 in FIG. 7).

  First, the path search unit 450 calculates the fitness of these gene codes (see step S702 in FIG. 7). In this calculation process, first, a logical topology corresponding to the node arrangement pattern indicated by the first gene code is virtually constructed, and then the shortest path between the nodes in the logical topology is calculated. The calculation of the shortest path can be performed using, for example, a known Dijkstra method. Subsequently, the path search unit 450 calculates, for each node, the amount of relay traffic when performing IP packet communication between nodes using each shortest path thus obtained. As described above, the relay traffic amount is stored in the objective function table 430 as an objective function. The path search unit 450 calculates the amount of relay traffic in the virtual logical topology based on the measurement result. Then, the path search unit 450 selects the largest amount of relay traffic (calculation result) of each node as the fitness. Similarity is calculated in the same manner for other gene codes. An example of the fitness calculated for each gene code is shown in FIG.

  Subsequently, the path search unit 450 selects the most excellent gene code from these 20 gene codes (see step S703 in FIG. 7). As described above, in this embodiment, the maximum amount of relay traffic of each node is the fitness. Therefore, it is considered that the concentration of traffic is less likely to occur as the fitness level is smaller. For this reason, in this embodiment, the gene code having the smallest fitness among the 20 gene codes is selected as the most excellent gene code. Hereinafter, the case where the gene code G2 in FIG. 9A is selected as the most excellent gene code will be described as an example.

  The path search unit 450 reads the most excellent gene code and determines whether or not the end condition is satisfied (see step S704 in FIG. 7). For example, when the fitness of the most excellent gene code falls below the threshold, the processing time of the genetic algorithm exceeds a preset processing time, or the computation processing for the genetic code is performed for a predetermined generation In this case, it is determined that the termination condition is satisfied. Here, it is assumed that the termination condition is not satisfied.

(3) Selection of Gene Code Next, the path search unit 450 selects a gene code (see step S705 in FIG. 7). “Selection” is a process of selecting an excellent gene from the genetic code, and corresponds to the fact that an excellent gene in the biological world tends to remain in a later generation. In the selection process, an excellent gene code, that is, a gene code having a low fitness level is selected more easily. As will be described below, in this embodiment, a predetermined probability is selected by assigning a selection probability such that the smaller the evaluation result is, the higher the probability of selection is to each gene code, and repeating a random selection a predetermined number of times. Select.

  In this selection process, 20 gene codes are selected using 19 gene codes G1, G3 to G20 excluding the most excellent gene code G2.

  First, the path search unit 450 assigns a selection probability to the gene codes G1, G3 to G20. Here, the selection probability is a probability that the gene code is selected when any one of the gene codes G1, G3 to G20 is selected. The selection probability of each gene code is determined so that an excellent gene code can be easily selected. As described above, in this embodiment, a genetic code with a lower fitness is better. Therefore, each selection probability is determined to be inversely proportional to the fitness. For example, the reciprocals for the respective fitness levels of the gene codes G1, G3 to G20 may be obtained, and these reciprocals may be weighted so that the total sum becomes “1”. Here, the selection probabilities of the gene codes G1, G3 to G20 are assumed to be x1, x3, x4,..., X20 (x1 + x3 + x4 +... + X20 = 1).

  Next, the path search unit 450 selects one of the gene codes G1, G3 to G20. First, the path search unit 450 causes the random number generation unit 440 to generate a random number R of 0 or more and less than 1, and further selects a gene code based on the value of the random number R. For example, the gene code G1 is selected when the random number R is 0 ≦ R <x1, the gene code G3 is selected when x1 ≦ R <x3, and the gene code G4 is selected when x3 ≦ R <x4. The According to such a method, any gene code can be selected with a probability corresponding to the above-mentioned selection probability.

  The path search unit 450 repeats the process of selecting any one of the 19 gene codes G1, G3 to G20 20 times. The selected gene code is included in the selection candidate gene code group in the subsequent selection process. Therefore, in this selection process, a case where the same gene code is selected more than once occurs once or more. It is also possible that a specific gene code is not selected even once. An example of the gene code selected by this selection process is shown in FIG. In FIG. 9B, the codes Gs1 to Gs20 are newly added to the respective gene codes after selection. The selected gene codes Gs1 to Gs20 are stored in the gene code table 420 together with the most excellent gene code G2 stored in advance.

(4) Crossover of gene code Next, the path search unit 450 crossovers the gene code (see steps S706 to S708 in FIG. 7). Crossing is a process of exchanging some elements randomly between genetic codes belonging to the same individual group, and corresponds to the fact that genes are crossed by crossing individuals in the biological world. As will be described below, in this embodiment, an arbitrary gene code pair is generated from the gene code, and randomly selected gene elements are exchanged between the gene code pairs.

  First, the path search unit 450 creates spare codes P1 to P20 as follows (see step S706 in FIG. 7).

  First, the path search unit 450 creates an order list C (0). In this embodiment, since the number of nodes of the IP backbone network 100 is 9, a sequence {1, 2, 3, 4, 5, 5 in which natural numbers 1 to 9 are arranged in ascending order as the ordered list C (0). 6, 7, 8, 9} is used. However, the arrangement order of the numerical values “1” to “9” is not limited to this.

  Next, the path search unit 450 reads the selected gene code Gs1 from the gene code table 420. In the example of FIG. 9B, the data structure of the gene code Gs1 is {1, 6, 3, 7, 4, 8, 9, 2, 5}.

  Subsequently, the path search unit 450 uses the order list C (0) and the gene code Gs1 to create the preliminary code P1 as follows (see FIG. 10).

  The path search unit 450 reads the first element from the gene code Gs1 = {1, 6, 3, 7, 4, 8, 9, 2, 5}. Here, the first element of the gene code Gs1 is “1”. Further, the path search unit 450 checks to which element in the order list C (0) this element “1” matches. Here, '1' is the first element of the ordered list C (0). Therefore, the path search unit 450 sets the first element of the backup code P1 to ‘1’ (see FIG. 10A).

  Next, the path search unit 450 removes the first element “1” from the order list C (0), thereby obtaining the order list C (1) = {2, 3, 4, 5, 6, 7, 8, 9} is created. Then, the path search unit 450 reads the second element of the gene code Gs1 and checks what element in the order list C (1) matches. Here, the second element of the genetic code Gs1 is “6”, and “6” is the fifth element of the ordered list C (1). Therefore, the path search unit 450 sets the second element of the spare code P1 to “5” (see FIG. 10B).

  Further, the path search unit 450 removes the fifth element “6” from the order list C (1), thereby obtaining the order list C (2) = {2, 3, 4, 5, 7, 8, 9}. create. Then, the path search unit 450 reads the third element of the gene code Gs1 and checks what element in the order list C (2) matches. Here, the third element of the genetic code Gs1 is “3”, and “3” is the second element of the ordered list C (2). Therefore, the path search unit 450 sets the third element of the backup code P1 to “2” (see FIG. 10C).

  Thereafter, the path search unit 450 repeats such an operation. In this way, the preliminary code P1 = {1, 5, 2, 4, 2, 3, 3, 1, 1} is finally obtained (see FIG. 10D).

  Subsequently, the path search unit 450 performs the same process for the gene code Gs2 = {3, 5, 1, 8, 6, 7, 4, 2, 9} (see FIG. 9B). Thereby, the spare code P2 = {3, 4, 1, 5, 3, 3, 2, 1, 1} can be obtained (see FIG. 10E).

  Similarly, preliminary codes P3 to P20 are created for the other gene codes Gs3 to Gs20.

  Next, the path search unit 450 divides the spare codes P1 to P20 into 10 pairs and performs crossover between the same pairs (see step S707 in FIG. 7). The pair determination method is arbitrary, and the combination may be determined using parameters. Here, a case where the pair of spare codes P1 and P2 is crossed will be described as an example (see FIG. 10F).

  In this crossover process, the path search unit 450 first determines the crossover position of the spare codes P1 and P2. The determination of the crossing position can be performed, for example, by causing the random number generation unit 440 to generate any one of natural numbers 1 to 9. For example, when the generated random number is '3', the path search unit 450 replaces the fourth and subsequent elements while keeping the third elements of the spare codes P1 and P2. As a result, Ph1 = {1, 5, 2, 5, 3, 3, 2, 1, 1} and Ph2 = {3, 4, 1, 4, 2, 3, 3, 1, 1} (see FIG. 10G).

  Subsequently, the path search unit 450 restores the crossed gene codes Gh1 and Gh2 from the crossed preliminary codes Ph1 and Ph2 as follows (see step S708 of FIG. 7, FIG. 11).

  First, the path search unit 450 reads the first element from the crossover spare code Ph1. Here, the first element of the spare code Ph1 is “1”. Further, the first element of the ordered list C (0) is “1”. Therefore, the path search unit 450 sets “1” as the first element of the gene code Gh1 (see FIG. 11A).

  Next, the path search unit 450 reads the second element of the backup code Ph1. Here, the second element of the spare code Ph1 is “5”, and the fifth element of the ordered list C (1) is “6”. Therefore, the path search unit 450 sets the second element of the gene code Gh1 to “6” (see FIG. 11B).

  Further, the path search unit 450 reads the third element of the spare code Ph1. Here, the third element of the spare code Ph1 is “2”, and the second element of the ordered list C (2) is “3”. Therefore, the path search unit 450 sets the third element of the gene code Gh1 to “3” (see FIG. 11C).

  Thereafter, the path search unit 450 repeats such an operation. Thus, the gene code Gh1 = {1, 6, 3, 8, 5, 7, 4, 2, 9} after crossover is finally obtained (see FIG. 11D).

  Subsequently, the path search unit 450 performs the same process for the spare code Ph2 = {3, 4, 1, 4, 2, 3, 3, 1, 1} after crossover (see FIG. 10G). . Thereby, the gene code Gh2 after crossover = {3, 5, 1, 7, 4, 8, 9, 2, 6} can be obtained (see FIG. 11E).

  Similarly, the gene codes Gh3 to Gh20 are restored from the other preliminary codes Ph3 to Ph20.

  As described above, the gene codes Gs1 to Gs20 are crossed, and as a result, the crossed gene codes Gh1 to Gh20 are obtained (see FIG. 9C).

(5) Mutation of gene code Next, the path search unit 450 generates a mutation in the gene code (see step S709 in FIG. 7). Mutation is a process of randomly rewriting a part of the elements constituting the gene code with a predetermined low probability, and corresponds to a sudden mutation of a gene in the biological world. By generating mutations, it becomes difficult for an arithmetic solution to fall into a local solution. As described below, in this embodiment, a part of or all of the genetic elements included in the genetic code is changed with a predetermined probability using a technique generally called inversion.

  The path search unit 450 stores the probability that a mutation will occur in advance. The occurrence probability of the mutation and the determination method thereof are arbitrary. For example, the occurrence probability may be changed using a parameter. Here, the probability of occurrence of mutation is 0.03.

  The path search unit 450 causes the random number generation unit 440 to generate random numbers R1 to R20 corresponding to the crossed gene codes Gh1 to Gh20, respectively. Here, each of the random numbers R1 to R20 is a real number not less than 0 and less than 1. Then, mutation is generated only for the gene code whose random numbers R1 to R20 are equal to or less than the set value of the mutation occurrence probability (here 0.03). FIG. 9D shows an example of random numbers R1 to R20. In the example of FIG. 9D, the gene code Gh1 is the target of the mutation process.

  In the mutation process, first, the random number generation unit 440 newly generates two random numbers h and k for each gene code that causes a mutation. These random numbers h and k are natural numbers of 1 to 9, respectively, and h ≠ k. Then, the path search unit 450 reverses the arrangement order of the gene code Gh1 from the h-th element to the k-th element. That is, when the gene code Gh1 before mutation is represented by the following formula (1) (here, h <k), the gene code Gm1 after mutation is represented by the following formula (2).

Gh1 = {s (1), ..., s (h-1), s (h), s (h + 1), ..., s (k-1), s (k), s (k + 1), ..., s (9)} ... (1)
Gm1 = {s (1), ..., s (h-1), s (k), s (k-1), ..., s (h + 1), s (h), s (k + 1), ..., s (9)} ... (2)
For example, when the random number generated corresponding to the gene code Gh1 is h = 2 and k = 8, Gh1 = {1, 3, 4, 2, 9, 8, 7, 6, 5} is Gm1 = {1 , 6, 7, 8, 9, 2, 4, 3, 5}.

  Next, the path search unit 450 randomly selects any one of the gene codes that have not been subjected to the mutation process. Then, the path search unit 450 rewrites the value of each element of the selected gene code to the value of each element of the most excellent gene code. As a result, the most excellent gene code is left in the next generation, and an optimal solution or a solution close to the optimal solution is easily obtained. FIG. 9E shows an example in which the gene code Gh20 is rewritten to the most excellent gene code G2.

  Thereafter, the path search unit 450 uses the gene codes in the individual group to perform the processing subsequent to step S702 again. The series of processes S702 to S710 are repeated until it is determined in step 704 that the end condition is satisfied.

  If it is determined in step S704 that the termination condition is satisfied, the path search unit 450 terminates the computation process by the genetic algorithm, and the best gene code at this time is the computation solution indicating the optimal logical topology. judge. Then, the path search unit 450 sends this calculation solution to the physical path allocation processing unit 460 (see FIG. 4).

  The physical path allocation processing unit 460 stores therein information related to the pair of end nodes of the current path and the physical path. The physical path allocation processing unit 460 compares the optimal solution received from the path search unit 450 with the stored information, and creates a mapping of the optical path with the matching end node. Then, the physical path allocation processing unit 460 allocates the physical path of the current path as it is to the physical path of the optimum optical path for the optical path with the matching end node. On the other hand, for the optical paths whose terminal nodes do not match, the shortest physical path is determined using, for example, a known Dijkstra method, and the determination result is assigned to the physical path of the optimal optical path. Here, when there are a plurality of types of physical paths having the same length, they may be randomly assigned. In this way, the physical path allocation processing unit 460 determines a terminal node pair and a physical route corresponding to the search result of the path search unit 450. The physical path information is stored in the physical path allocation processing unit 460 as current path information, and is sent to the optical path setting management device 230 (see FIG. 2) via the transmission processing unit 470.

  The optical path setting management device 230 changes the physical route of the IP backbone network 100 by controlling the optical switch control device 371 (see FIG. 3) in each of the nodes N1 to N9 using the received physical route information. Further, the optical path setting management device 230 changes a routing table (not shown) stored in the IP router in each of the nodes N1 to N9 according to the physical route information.

  As described above, according to the optimum optical path arrangement search device according to this embodiment, the optimum cut-through path can be searched by a simple process using a genetic algorithm.

Second Embodiment Next, a second embodiment of the present invention will be described with reference to FIGS. This embodiment is an example in which the optimum path location search apparatus according to the present invention is applied to an IP backbone network that uses the route location transition method (Japanese Patent Application No. 2004-358367) already proposed by the present inventors.

  The data plane configuration and control plane configuration of the IP backbone network according to this embodiment are the same as in the case of the first embodiment (see FIGS. 1 and 2). The configurations of the nodes N1 to N9, the traffic measurement management device 210, and the optimum optical path arrangement search device 220 are the same as those in the first embodiment (see FIGS. 3 and 4). On the other hand, the optical path setting management apparatus 230 of this embodiment uses the physical path information received from the optimum optical path arrangement search apparatus and uses the above-described path arrangement transition method (Japanese Patent Application No. 2004-358367) to create an IP backbone network. Change the optical path setting.

  FIG. 12 is a conceptual diagram showing a logical topology of the IP backbone network according to this embodiment. As can be seen from FIG. 12, in this embodiment, a logical topology of a ring structure is constructed on the IP backbone network by setting a total of nine logical paths between the logical arrangements L1 to L9.

  FIG. 13 shows an example in which nodes N1 to N9 are arranged in the logical arrangements L1 to L9 in FIG. Similar to the first embodiment described above, in this embodiment as well, an arbitrary node can be arranged in each of the logical arrangements L1 to L9 by appropriately setting a normal optical path or cut-through path. The search for the optimum optical path uses the same genetic algorithm as in the first embodiment.

  Also in the IP backbone network according to this embodiment, as in the first embodiment, for example, carrier waves of four types of wavelengths are multiplexed by WDM to transmit / receive IP packets (see FIG. 3). However, in this embodiment, each node uses a carrier wave of the same wavelength for the same logical path. In the first embodiment described above, for example, the received carrier wave having the wavelength λ1 is converted into the wavelength λ0 by the optical transceiver 341 and routing is performed, and then the wavelength λ2, at any of the other optical transceivers 342 to 344, is performed. In some cases, the signal is re-converted into a carrier wave of λ3 or λ4. On the other hand, in this embodiment, for example, an IP packet received using a carrier wave of wavelength λ1 is always reconverted to a carrier wave of wavelength λ1 and transmitted.

  As shown in FIGS. 12 and 13, all the nodes N1 to N9 are used as termination nodes for two paths. Therefore, in each of the nodes N1 to N9, there are two types of carrier wavelengths that the IP router 351 relays (one wavelength used for relay and one wavelength used for cut-through). On the other hand, each IP router 351 can use four types of carrier wavelengths (see FIG. 3). For this reason, in each of the nodes N1 to N9, two types of wavelengths among the four types of assigned wavelengths are used for the construction of the working logical path, and the remaining two types of wavelengths are reserved.

  Thus, according to this embodiment, it is possible to connect all the nodes N1 to N9 using fewer cut-through paths than in the first embodiment. That is, in this embodiment, it is possible to construct a logical topology using fewer types of wavelengths than in the first embodiment. Therefore, it is possible to secure the above-described spare optical path while realizing an optimal optical path arrangement. The spare optical path can be used for relief when sudden congestion / failure occurs in the working path.

  Also in the IP backbone network according to this embodiment, an optimal optical path is searched using the same genetic algorithm (see FIGS. 7 to 11) as the IP backbone network of the first embodiment. FIG. 14 conceptually shows the configuration of the gene code corresponding to the node arrangement of FIG. That is, according to the optimum optical path arrangement search device according to this embodiment, in an IP backbone network having a ring structure having a current logical path and a backup logical path, an optimum cut is performed by simple processing using a genetic algorithm. Through pass can be searched.

Third Embodiment Next, a third embodiment of the present invention will be described with reference to FIGS. The optimum path arrangement search apparatus according to this embodiment searches the optimum optical path arrangement in consideration of not only the maximum value of relay traffic but also the maximum value of traffic per carrier. This is different from the optimum path arrangement search device of the first embodiment.

  The data plane configuration and control plane configuration of the IP backbone network according to this embodiment are the same as in the case of the first embodiment (see FIGS. 1 and 2). The configurations of the nodes N1 to N9, the traffic measurement management device 210, and the optical path setting management device 230 are the same as those in the first embodiment (see FIGS. 2 and 3).

  FIG. 15 is a block diagram showing an internal configuration of the optimum optical path arrangement search device. In FIG. 15, the components given the same reference numerals as those in FIG. 4 are the same as those in FIG. 4.

  As can be seen from FIG. 15, the optimum optical path arrangement search device according to this embodiment includes an objective function / fitness conversion unit 1510. The objective function / fitness conversion unit 1510 calculates the superiority order of the gene code from the maximum value of the relay traffic amount of each gene code and the maximum value of the IP packet transmission amount per carrier wave (described later).

  Hereinafter, the calculation operation of the optimum optical path arrangement search device 220 will be described with reference to FIGS. 16 and 17.

  FIG. 16 is a flowchart schematically showing a calculation process of the optimum optical path arrangement search device according to this embodiment.

(1) Generation of initial population First, as in the first embodiment, the random number generation unit 440 (see FIG. 15) generates an initial population consisting of a plurality of gene codes (see step S1601 in FIG. 16). . In this embodiment as well, as in the first embodiment, the initial population has 20 gene codes, and each gene code has 9 elements. The generated gene code is stored in the gene code table 420.

(2) Evaluation of gene code Next, the path search unit 450 evaluates each gene code belonging to the individual group as follows (see steps S1602 and S1603 in FIG. 16).

  First, the path search unit 450 calculates the fitness of these gene codes (see step S1602 in FIG. 16). In the fitness calculation processing of this embodiment, first, similarly to the first embodiment, a logical topology corresponding to the node arrangement pattern is virtually constructed for each gene code, and each node in the logical topology is further constructed. The shortest path between each is calculated. Subsequently, the path search unit 450 calculates a relay traffic amount and a traffic amount per carrier for each of the shortest paths for each node. The relay traffic amount and the traffic amount per carrier wave are stored in the objective function table 430 as an objective function. Then, the objective function / fitness conversion unit 1510 calculates the superiority ranking of each gene code using the largest one of the relay traffic amounts and the largest one of the traffic amounts per carrier wave.

  FIG. 17 is a graph for explaining a method of calculating the superiority ranking. The horizontal axis f is the maximum value of the relay traffic amount, and the vertical axis g is the maximum value of the traffic amount per carrier wave. The objective function / fitness conversion unit 1510 receives from the objective function table 430 the maximum value of the relay traffic amount and the maximum value of the traffic amount per carrier corresponding to each gene code. Then, for each gene code, the number of gene codes in which both the maximum value of the relay traffic amount and the maximum value of the traffic amount per carrier wave are smaller than the gene code is counted. For example, in the case of the gene code C7 of FIG. 17, both the maximum value of the relay traffic amount (ie, the f coordinate) and the maximum value of the traffic amount per carrier wave (ie, the g coordinate) are smaller than the gene code C7. There are only three gene codes C1, C3 and C4. Therefore, the superiority ranking of the genetic code C7 is “4”. Regarding the gene codes C2 and C4, only the gene code C1 has both the f coordinate and the g coordinate smaller than the gene code. Accordingly, the superiority ranking of the gene codes C2 and C4 is “2”. Furthermore, since the gene codes C1, C3, C5, and C6 do not have a gene code having both small f-coordinates and g-coordinates, the superiority rank is “1”.

  Subsequently, the path search unit 450 selects the most excellent gene code from the 20 gene codes (see step S1603 in FIG. 16). In this embodiment, the gene code with the lowest superiority rank is the best gene code.

  The path search unit 450 reads the most excellent gene code and determines whether or not the end condition is satisfied (see step S1604 in FIG. 16). If there are several best genetic codes, you can select one of them at random, check the termination conditions for all the best genetic codes, or use other methods. May be. In this embodiment, when the maximum value of the relay traffic amount of the most excellent gene code and the maximum value of the traffic amount per carrier wave are both equal to or less than the threshold value, the processing time of the genetic algorithm is set to a preset processing time. When the number of times is exceeded or when the calculation process for the gene code is performed for a predetermined generation, it is determined that the termination condition is satisfied.

(3) Selection of gene code When the termination condition is not satisfied, the path search unit 450 selects a gene code as follows (see step S1605 in FIG. 16).

  In the selection process of this embodiment, the gene code with the superiority ranking “1” is selected unconditionally. Then, for example, 20 gene codes are selected from the remaining gene codes. Also in this embodiment, as in the first embodiment, the path search unit 450 assigns a selection probability to a target gene code (a gene code whose superiority rank is not “1”). This selection probability is assigned so that the gene code with the lower priority rank becomes higher. Then, the path search unit 450 selects a gene code 20 times, for example, in the same manner as in the first embodiment. Thereby, a predetermined number of gene codes are selected with a probability corresponding to the above-mentioned selection probability. The selected gene code is stored in the gene code table 420.

(4) Crossover of gene code Next, the path search unit 450 creates a spare code (see step S1606 in FIG. 16) in the same manner as in the first embodiment, and crosses over the spare code (see FIG. 16). Further, the gene code is restored from the preparatory code after the crossover (see step S1608 in FIG. 16).

(5) Mutation of gene code Furthermore, the path search unit 450 generates a mutation in the gene code in the same manner as in the first embodiment (see step S1609 in FIG. 16).

  Thereafter, the path search unit 450 uses the gene codes in the individual group to perform the processing subsequent to step S1602 again. The series of processes S1602 to S1609 are repeated until it is determined in step 1604 that the end condition is satisfied.

  If it is determined in step S1604 that the termination condition is satisfied, the path search unit 450 terminates the arithmetic processing by the genetic algorithm, and the best gene code at this time is an arithmetic solution indicating an optimal logical topology. judge. Then, the path search unit 450 sends this calculation solution to the physical path allocation processing unit 460 (see FIG. 15).

  In this embodiment, instead of selecting all the gene codes having the superiority rank “1” in the above selection process (see step S1605 in FIG. 16), the most excellent gene code is not introduced in the mutation process. It was. However, as in the first embodiment described above, a gene code other than the most excellent gene code may be used in the selection process, and the most excellent gene code may be introduced in the mutation process.

  Other operations are the same as those in the first embodiment, and a description thereof will be omitted.

  As described above, according to the optimum optical path arrangement search device according to this embodiment, an optimum cut-through path is searched using a genetic algorithm considering both the amount of relay traffic and the amount of traffic per carrier wave. can do.

It is a conceptual diagram which shows the data plane structure of the IP backbone network which concerns on 1st Embodiment. It is a conceptual diagram which shows the control plane structure of the IP backbone network which concerns on 1st Embodiment. It is a block diagram which shows the internal structure of the node which concerns on 1st Embodiment. It is a block diagram which shows the internal structure of the optimal optical path arrangement | positioning search apparatus which concerns on 1st Embodiment. It is a conceptual diagram which shows the logical topology of the IP backbone network which concerns on 1st Embodiment. It is a conceptual diagram which shows an example of the node arrangement | positioning to the logical topology shown in FIG. It is a flowchart which shows roughly the arithmetic processing of the optimal optical path arrangement | positioning search apparatus which concerns on 1st Embodiment. It is a table | surface which shows the data structure of the gene code used in 1st Embodiment. (A)-(E) is a table | surface for demonstrating operation | movement of the optimal optical path arrangement | positioning search apparatus which concerns on 1st Embodiment. (A)-(G) is a table | surface for demonstrating operation | movement of the optimal optical path arrangement | positioning search apparatus which concerns on 1st Embodiment. (A)-(E) is a table | surface for demonstrating operation | movement of the optimal optical path arrangement | positioning search apparatus which concerns on 1st Embodiment. It is a conceptual diagram which shows the logical topology of the IP backbone network which concerns on 2nd Embodiment. It is a conceptual diagram which shows an example of the node arrangement | positioning to the logical topology shown in FIG. It is a table | surface which shows the data structure of the gene code used in 2nd Embodiment. It is a block diagram which shows the internal structure of the optimal optical path arrangement | positioning search apparatus which concerns on 3rd Embodiment. It is a flowchart which shows roughly the arithmetic processing of the optimal optical path arrangement | positioning search apparatus which concerns on 3rd Embodiment. It is a table | surface for demonstrating operation | movement of the optimal optical path arrangement | positioning search apparatus which concerns on 3rd Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 IP backbone network N1-N9 node F1-F24 Optical fiber 210 Traffic measurement management apparatus 220 Optimal optical path arrangement search apparatus 230 Optical path setting management apparatus 240,250 Control network 311-314 Optical amplifier 321-324 Optical multiplexer / demultiplexer 331, 332 Optical switch 341-344 Optical transceiver 351 IP router 361, 362 Traffic measurement device 371 Optical switch control device 410 Reception processing unit 420 Gene code table 430 Objective function table 440 Random number generation unit 450 Path search unit 460 Physical path allocation processing unit 470 Transmission processor

Claims (20)

An optimum path arrangement search device for searching a path arrangement with a small amount of traffic using a plurality of gene codes indicating a node arrangement on a logical topology,
Evaluation means for evaluating traffic of the logical topology indicated by each of the gene codes;
A selection that selects a predetermined number of gene codes by assigning a selection probability such that the smaller the evaluation result is, to each of the gene codes, and repeating the selection according to the selection probability a predetermined number of times. Means,
Crossover means for generating an arbitrary gene code pair from the gene code and exchanging randomly selected gene elements between the gene code pairs;
Mutation means for changing a part or all of the genetic elements included in the genetic code with a predetermined probability;
An optimum path arrangement search device comprising:   2. The optimum path arrangement search device according to claim 1, wherein the first gene code is generated by random number generation means.   The evaluation means calculates the relay traffic of the logical topology indicated by the gene code for each node of the logical topology, and sets the largest relay traffic as the evaluation result of the gene code. 2. The optimum path arrangement search device according to 2.   The evaluation unit calculates the relay traffic of the logical topology indicated by the gene code and the traffic per carrier for each node of the logical topology, and selects the largest relay traffic and the traffic per carrier, respectively. The number of the gene codes in which both the largest relay traffic and traffic per carrier wave are smaller than the gene code are counted, and a value obtained by adding '1' to the number is used as an evaluation result. Item 3. The optimum path arrangement search device according to Item 1 or 2.   After the selection means assigns a selection probability that is inversely proportional to the result of the evaluation to each of the gene codes, selection according to the selection probability without removing the already selected gene code from the selection target 5 is repeated a predetermined number of times, and the optimum path arrangement search device according to any one of claims 1 to 4.   The crossover means generates an ordered list in which elements having the same value as the genetic elements of the genetic code are randomly arranged, and rearranges the arrangement order of the genetic elements in each genetic code in correspondence with the ordered list. A plurality of spare codes, and by dividing the plurality of spare codes into two, a plurality of spare code pairs are generated, and the gene elements to be exchanged are randomly assigned to each spare code pair. The sequence of the gene elements is restored in correspondence with the order list after the determination and exchanging the gene elements between the spare code pairs based on the determination. 5. The optimum path arrangement search device according to any one of 5).   The mutation means determines the occurrence / non-occurrence of a mutation in the gene code by comparing a random number generated for each gene code with a predetermined threshold, and 2 for each gene code for generating a mutation. Random numbers h, k (1 ≦ h, k ≦ n; n is the number of the gene elements included in the gene code) are generated, and the gene elements from the h-th to the k-th included in the gene code are generated. 7. The optimum path arrangement search apparatus according to claim 1, wherein the arrangement order is reversed. The evaluation means evaluates the m gene codes;
The selection means selects m gene codes from m-1 gene codes excluding the gene code with the smallest evaluation result;
The mutation means replaces any of the genetic codes that did not cause a mutation with a genetic code having the smallest evaluation result;
The optimal path arrangement search device according to any one of claims 1 to 7,   The optimal path arrangement search device according to claim 1, wherein the logical topology has a torus structure.   The optimal path arrangement search device according to claim 1, wherein the logical topology has a ring structure. An optimal path arrangement search method for searching a path arrangement with a small amount of traffic using a plurality of gene codes indicating a node arrangement on a logical topology,
An evaluation step of evaluating traffic of the logical topology indicated by the genetic code;
By assigning a selection probability such that the smaller the evaluation result in the evaluation step is to each gene code, and repeating the selection according to the selection probability a predetermined number of times, a predetermined number of the gene codes are A selection step to select;
A crossover step of generating an arbitrary gene code pair from the gene code selected in the selection step, and exchanging randomly selected gene elements between the gene code pairs;
A mutation step of changing a part or all of the genetic elements included in the genetic code crossed in the crossover step with a predetermined probability;
The optimal path arrangement | positioning search method characterized by providing.   The optimal path arrangement search method according to claim 11, further comprising a gene code generation step of generating the first gene code by generating a random number.   The evaluation step is a step of calculating relay traffic of the logical topology indicated by the gene code for each node of the logical topology, and setting the largest relay traffic as an evaluation result of the gene code. Item 13. The optimal path arrangement search method according to Item 11 or 12.   The evaluation step calculates the relay traffic of the logical topology indicated by the genetic code and the traffic per carrier for each node of the logical topology, and selects the largest relay traffic and the traffic per carrier, respectively. The largest relay traffic and the traffic per carrier are both counted by counting the number of the gene codes that are smaller than the gene code, and a value obtained by adding '1' to the number is used as an evaluation result. The optimal path arrangement search method according to claim 11 or 12.   After the selection step assigns a selection probability that is inversely proportional to the result of the evaluation to each of the gene codes, the selection according to the selection probability is performed without removing the already selected gene code from the selection target. The optimal path arrangement search method according to claim 11, wherein the step is a predetermined number of times.   The crossover step generates an ordered list in which elements having the same value as the gene elements of the gene code are randomly arranged, and rearranges the arrangement order of the gene elements in each of the gene codes in correspondence with the ordered list. A plurality of spare codes, and by dividing the plurality of spare codes into two, a plurality of spare code pairs are generated, and the gene elements to be exchanged are randomly assigned to each spare code pair. And determining the sequence of the gene elements in correspondence with the order list after exchanging the gene elements between the spare code pairs based on the determination. Item 16. The optimal path arrangement search method according to any one of Items 11 to 15.   The mutation step determines the occurrence / non-occurrence of a mutation in the gene code by comparing a random number generated for each gene code with a predetermined threshold, and 2 for each gene code for generating a mutation. Random numbers h, k (1 ≦ h, k ≦ n; n is the number of the gene elements included in the gene code) are generated, and the gene elements from the h-th to the k-th included in the gene code are generated. The optimal path arrangement search method according to any one of claims 11 to 16, which is a step of reversing the arrangement order. The evaluation step is a step of evaluating the m gene codes;
The selection step is a step of selecting m gene codes from m-1 gene codes excluding the gene code having the smallest evaluation result;
The mutation step includes a process of replacing any of the genetic codes that did not cause mutation with a genetic code having the smallest evaluation result.
The optimal path arrangement search method according to claim 11, wherein:   The optimal path arrangement search method according to claim 11, wherein the logical topology has a torus structure.   The optimal path arrangement search method according to claim 11, wherein the logical topology has a ring structure.

Images