JPWO2012128352A1 - Local information propagation method, network, computation node, node and program - Google Patents

Abstract

To provide a method for rapidly propagating local information to a required range. Maximizing the maximum eigenvalue of an adjacency matrix representing the partial network so that the partial network propagating local information generated in the network has a predetermined size, and the second minimum eigenvalue of the normalized Laplacian matrix of the partial network Calculating the configuration of the partial network that maximizes the network, notifying the calculated configuration information to all nodes in the partial network, and if local information occurs, the configuration information is used. And propagating the local information to the partial network.

Description

[Description of related applications]
The present invention is based on the priority claim of Japanese Patent Application No. 2011-065776 (filed on Mar. 24, 2011), the entire contents of which are incorporated herein by reference. Shall.

  The present invention relates to a local information propagation method, a network, a computation node, a node, and a program, and more particularly to a local information propagation method, a network, and a computation node for propagating local information such as a failure or a failure that has occurred in a network or system. , Nodes and programs.

  When an element constituting a network or system has a failure or failure, the failure / failure information is collected, the cause is identified, and control is performed. In this case, it is information collection and cause location identification that have the largest load and time. In other words. Performing these two processes at high speed is important in controlling the network and system.

  In order to perform the above information collection and cause location identification at high speed, information that the failure or failure has occurred in the control area or control node responsible for processing such as information collection or cause location identification when a failure or failure occurs Must be propagated at high speed.

  Non-patent document 1 is one of prior arts of the present invention. This method uses TCP / IP (Transmission Control Protocol / Internet Protocol), which is a communication protocol generally used on the Internet, to determine failure information in the network in advance for each router in the network. By causing failure information to be propagated in order according to the route, each router is recalculated. Other prior art includes Non-Patent Document 2. Although there are various variations in this method, the purpose is to determine a detour that does not pass through a failure / failure location faster than Non-Patent Document 1.

  Non-Patent Documents 3 to 8 are documents that are other background arts of the present invention. The relationship with the present invention will be described in detail later.

“The Addition of Explict Notification Notification (ECN) to IP”, RFC 3168. Minas Gjoka and two others, “Evaluation of IP Fast Route Proposals”, [online], [searched on March 1, 2011], Internet <URL: http://www.Minasgjoka.com/papers/ipfrr06 .pdf) Akira Namaten, “Group Intelligence and Networked Control”, Department of Information Engineering, National Defense University, [online], [Searched on March 1, 2011], Internet <URL: http://www.nda.ac.jp /~nama/Top/InvitedTalk/10-11-30.pdf> Luca Donetti and two others, “Optimal network topologies: Expanders, Cages, Ranuman graphs, Ent ed networks and all tat.”, Journal of Tart. Lester Ingber, “Simulated annealing: Practice versus theory”, [online], [searched on March 1, 2011], Internet <URL: http://www.ingber.com/asa93_sapvt.pdf> Saumitra M. et al. Das et al., “Performance Comparison of Scalable Location Services for Graphical Ad Hoc Routing”, IEEE INFOCOM2005, pp. 1228-1239 IBM, “TCP / IP Tutorial and Technical Description”, [online], [searched on March 1, 2011], Internet <URL: http://www-06.ibm.com/jp/ support / redbooks / TCP_IP / GG88400500.pdf> "Spanning Tree", [online], [Search on March 1, 2011], Internet <URL: http://en.wikipedia.org/wiki/%E3%82%B9%E3%83% 91% E3% 83% 8B% E3% 83% B3% E3% 82% B0% E6% 9C% A8>

  The following analysis is given by the present invention.

  In the technique described in Non-Patent Document 1, there is no restriction on the propagation area of failure / failure information, but the failure / failure information is notified to the entire network using a normal routing protocol, and the router from which the information arrived Since route recalculation is performed in order, the arrival time of information is slow, so that there is a problem that control takes time.

  Similarly, the technology described in Non-Patent Document 2 is not limited in the propagation area of failure / failure information, and the failure / failure information is notified to the entire network according to the route configured by the normal route control protocol, and the information arrives. Since the route is recalculated in order from the router, the arrival time of the information is slow, so that control takes time.

  The present invention has been made in view of the above-described circumstances, and its object is to make the propagation area of local information (hereinafter referred to as “local information”) in the network as wide and necessary as possible. It is an object to provide a method for restricting a part and propagating the information at high speed.

  According to the first aspect of the present invention, the maximum eigenvalue of an adjacency matrix representing the partial network is maximized so that the partial network for propagating local information generated in the network has a predetermined size, and the portion Calculating a configuration of a partial network that maximizes a second minimum eigenvalue of a normalized Laplacian matrix of the network, notifying all of the nodes in the partial network of the calculated configuration information, and local When information is generated, a method of propagating local information is provided, including the step of propagating the local information to the partial network using the configuration information. This method is linked to a specific machine, which is a calculation node that is arranged in a network and determines the propagation range of the local information.

  According to the second aspect of the present invention, as a partial network for propagating local information generated in a network, the maximum eigenvalue of an adjacency matrix representing the network is maximized, and the normalized Laplacian matrix of the network is A calculation node comprising: means for calculating a configuration of a partial network that maximizes a minimum eigenvalue; and means for notifying all of the nodes in the partial network of the calculated configuration information. When information is generated, a network is provided in which nodes belonging to the partial network propagate the local information using a protocol defined for the propagation of the local information.

  According to the third aspect of the present invention, as a partial network for propagating local information generated in a network, the maximum eigenvalue of an adjacency matrix representing the network is maximized, and the normalized Laplacian matrix of the network is 2. A computing node is provided comprising means for calculating a configuration of a partial network that maximizes a minimum eigenvalue, and means for notifying all of the nodes in the partial network of the calculated configuration information. .

  According to the fourth aspect of the present invention, when local information is generated, a node that propagates the local information using a predetermined protocol is provided using configuration information communicated from the above-described calculation node. .

  According to the fifth aspect of the present invention, the maximum eigenvalue of an adjacency matrix representing the partial network is maximized so that the partial network for propagating local information generated in the network has a predetermined size, and the portion A process of calculating the configuration of the partial network that maximizes the second minimum eigenvalue of the normalized Laplacian matrix of the network, and a process of notifying all the nodes in the partial network of the calculated configuration information. A program to be executed by a computing node connected to the network is provided. This program can be recorded on a computer-readable storage medium. That is, the present invention can be embodied as a computer program product.

  According to the present invention, it is possible to propagate local information to a necessary range at high speed.

It is a figure which shows the structure of the network of the 1st Embodiment of this invention. It is a figure which shows the structure of the calculation server of the 1st Embodiment of this invention. It is a flowchart for demonstrating the propagation method (preparation stage) of the local information of the 1st Embodiment of this invention. It is a figure for demonstrating the calculation process in the calculation server of the 1st Embodiment of this invention. It is another figure for demonstrating the calculation process in the calculation server of the 1st Embodiment of this invention. It is a flowchart for demonstrating the propagation method (propagation step) of the local information of the 1st Embodiment of this invention. It is a flowchart for demonstrating the propagation method (preparation stage) of the local information of the 2nd Embodiment of this invention. It is a flowchart for demonstrating the propagation method (propagation step) of the local information of the 2nd Embodiment of this invention. It is a flowchart for demonstrating the propagation method (preparation stage) of the local information of the 3rd Embodiment of this invention. It is a flowchart for demonstrating the propagation method (propagation step) of the local information of the 3rd Embodiment of this invention. It is a flowchart for demonstrating the propagation method (preparation stage) of the local information of the 4th Embodiment of this invention. It is a flowchart for demonstrating the propagation method (propagation step) of the local information of the 4th Embodiment of this invention.

  First, two mathematical properties relating to the eigenvalues of the network (hereinafter referred to as “NW”) that form the basis of the present invention will be described.

(Characteristic 1) In order to propagate local information generated in an NW to the widest possible area, it is necessary to configure an NW that increases the maximum eigenvalue of the adjacent matrix of the NW (Non-patent Document 3). reference).

(Characteristic 2) In order to propagate the local information generated in the NW as wide as possible as fast as possible, follow the Markov chain between any two nodes s and t in the NW given by the following equation [Formula 1] It is necessary to construct an NW that minimizes the first walk time time of the random walk (Non-patent Document 4: top formula on page 5), that is, an NW that maximizes the second minimum eigenvalue of the normalized Laplacian matrix of the NW. It is.

  Here, M is the total number of links in the NW, and d (s) and d (t) are the orders of the nodes s and t, respectively. Also, 0 ≦ λ_v ≦ 2 is the vth (v is a natural number of 2 or more) eigenvalue from the smaller NW normalized Laplacian matrix, and u_ (l, s) is l from the smaller NW normalized Laplacian matrix. It is the value of the sth row from the top of the eigenvector having a magnitude of 1 with respect to the first eigenvalue (l is a natural number of 2 or more).

(Property 3) An NW satisfying the above (Property 1) and (Property 2) is an NW that is uniform in the order of each node in the NW, betweenness centrality, average distance, minimum loop size, and order correlation distribution. It is. For example, the Entangled NW described in Non-Patent Document 4 is one of them.

  Due to the above properties, the two conflicting areas of restricting the propagation area of local information generated in the NW (or system) to the widest possible area and the necessary information and propagating the failure / failure information at high speed are as follows. In order to realize the object, it is derived that the NW should be configured such that the second minimum eigenvalue of the NW normalized Laplacian matrix is increased and the maximum eigenvalue of the adjacent matrix of NW is increased.

  For example, when a local event occurs in an NW composed of nodes and links, the local information can be efficiently transmitted by performing the following preparation. Here, the local information includes the degree of the event, the identification information of the node where the event has occurred, the time of occurrence, and the like. Further, it is assumed that a node in the NW has an NW control function for performing control prescribed in advance by an NW administrator or the like when this local information is received.

(Procedure 1) A communication protocol for notifying the local information is prepared. The communication protocol for transmitting local information is preferably a dedicated communication protocol that can communicate at a higher speed than the existing routing control protocol.

(Procedure 2) A temporary local information notification NW range for propagating local information is determined. Also, the eigenvalue calculation node is determined from the communication nodes belonging to the NW range. The “NW range for temporary local information notification” determined here is a range in which the NW control function possessed by all communication nodes belonging to the NW can recognize the arrival time of the local information as the same time. That is.

(Procedure 3) In the eigenvalue calculation node, an NW range adjacency matrix and a normalized Laplacian matrix for temporary local information notification are generated.

(Procedure 4) In the eigenvalue calculation node, the maximum eigenvalue of the adjacent matrix in the NW range for temporary local information notification and the second minimum eigenvalue of the normalized Laplacian matrix are obtained.

(Procedure 5) While appropriately adding and deleting the number of communication nodes and the number of links in the temporary local information notification NW range by an arbitrary optimization algorithm in the eigenvalue calculation node (Procedure 3) By repeating (Procedure 4), the NW range for local notification that maximizes the maximum eigenvalue of the adjacency matrix and the second minimum eigenvalue of the normalized Laplacian matrix is determined.

(Procedure 6) The configuration information of the local information notification NW is transmitted from the eigenvalue calculation node to the nodes belonging to the NW range determined in (Procedure 5). The NW configuration information can be transmitted using any communication protocol. Of course, the communication protocol determined in (Procedure 1) can also be used.

(Procedure 7) (Procedure 2) to (Procedure 6) are performed for the remaining NWs outside the NW range determined in (Procedure 6) in the communication NW. Eventually, the entire NW is divided into several NW ranges (partial NWs) for notification of local information.

(Procedure 8) When local information occurs in the NW range for notification of local information, the local information is transmitted by the communication protocol prepared in (Procedure 1), starting from the node where the failure has occurred.

(Procedure 9) Upon receiving the local information, the node belonging to the NW range for notifying the local information transmits the local information to other adjacent nodes and performs NW control corresponding to the received local information.

  As described above, it is possible to control the partial NW at high speed by transmitting local information to a necessary range at high speed. Of course, if the local information is transmitted simultaneously and frequently to each part of the partial NW, the entire NW can be controlled globally.

  Further, when a dedicated communication protocol capable of communicating at higher speed than the existing route control protocol is used as the communication protocol prepared in (Procedure 1), the local information transmission NW independent of the route control protocol NW is used. In addition, it is possible to perform high-speed NW control by using a communication protocol dedicated to NW.

[First Embodiment]
Next, a first embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 is a diagram showing a network configuration according to the first embodiment of this invention. Referring to FIG. 1, a communication network (hereinafter referred to as “communication NW1”) configured by a plurality of NWs (W1_x to W9_x) for notification of local information is shown.

  The local information notification NWs (W1_x to W9_x) are respectively configured by communication nodes (see CN in FIG. 1). In addition, a calculation server (see CS in FIG. 1) that calculates the NW range for notification of local information is arranged everywhere in the communication NW1. Note that some of the communication nodes CN may have a function as a calculation server (see CS in FIG. 1).

  FIG. 2 is a diagram illustrating a configuration of the calculation server CS according to the first embodiment of this invention. Referring to FIG. 2, the communication unit 11, the Laplacian matrix calculation unit 12, the second minimum eigenvalue calculation unit 13, the adjacency matrix calculation unit 14, the maximum eigenvalue calculation unit 15, and the first arrival time calculation unit 16 are compared. The structure provided with the part 17 and the NW structure determination part 18 is shown.

  The communication unit 11 is means for exchanging various information with the communication node CN. Specifically, the configuration information of the communication NW1 and the configuration information of the local information notification NW are transmitted and received.

  The Laplacian matrix calculation unit 12 generates a standardized Laplacian matrix NL in the NW range instructed from the NW configuration determination unit 18 based on the configuration information of the communication NW1 received from the communication node CN.

  The second minimum eigenvalue calculation unit 13 calculates the second minimum eigenvalue of the normalized Laplacian matrix NL generated by the Laplacian matrix calculation unit 12 and outputs it to the first arrival time calculation unit 16.

  The adjacency matrix calculation unit 14 generates an adjacency matrix A in the NW range instructed from the NW configuration determination unit 18 based on the configuration information of the communication NW1 received from the communication node CN.

  The maximum eigenvalue calculation unit 15 calculates the maximum eigenvalue of the adjacency matrix A generated by the adjacency matrix calculation unit 14 and outputs the maximum eigenvalue to the first arrival time calculation unit 16.

  The first arrival time calculation unit 16 substitutes the calculation result received from the second minimum eigenvalue calculation unit 13 and the maximum eigenvalue calculation unit 15 and the parameter instructed from the NW configuration determination unit 18 into the above [Equation 1]. To calculate the first arrival time.

  The comparison unit 17 is a means for comparing the first arrival time calculated by the first arrival time calculation unit 16 and Nmax determined by the NW configuration determination unit 18.

  The NW configuration determination unit 18 calculates Nmax indicating the maximum number of physical links that can transmit failure / failure information i, which will be described later, and calculates the above until the first arrival time by the first arrival time calculation unit 16 is less than Nmax. It is means for controlling each part of the server CS, repeating the calculation, and determining the range and configuration of the local information notification NW.

  Note that each unit (processing means) of the calculation server shown in FIG. 2 can also be realized by a computer program that causes a computer constituting the calculation server to execute the above-described processes using its hardware.

  Next, the operation of this embodiment will be described in detail with reference to the drawings. FIG. 3 is a flowchart for explaining a local information propagation method (preparation stage) according to the first embodiment of this invention. Hereinafter, it is assumed that the communication NW1 has already been configured and the logical network NWL1 has been determined by path control. A procedure for setting a propagation area in a wide range and necessary portion of local failure / failure information generated in the communication NW1 and propagating at high speed will be described. In the following description, the failure / failure information includes the degree of failure / failure, the identification information of the node where the failure / failure has occurred, the time of occurrence, and the like.

(Procedure S1-1) When a failure / failure occurs, a communication protocol P2 for transmitting failure / failure information i is prepared. As this communication protocol P2, a dedicated communication protocol capable of communicating at higher speed than the route control protocol P1 is selected. More preferably, a communication protocol suitable for spreading the failure / failure information i is selected using a route on an arbitrary part NW (local information notification NW) of the communication NW.

(Procedure S1-2) Next, assuming that n = 1, a temporary failure that propagates the failure / failure information i of the communication NW as in (Procedure S1-2-1) to (Procedure S1-2-5) below A range W1_n for failure information notification is determined. Here, the range W1_n is a range in which the NW control function possessed by all communication nodes belonging to the range W1_n can recognize the time when the failure / failure information i arrives as the same time. Hereinafter, in this range W1_n, W1_x that satisfies the above (property 1) and (property 2) is obtained, and the local information notification NW is configured.

(Procedure S1-2-1) First, one calculation server CS for calculating eigenvalues connected to an arbitrary communication node CN not belonging to the configured local information notification NW of the communication NW1 is determined. FIG. 4 is a diagram showing a state in which one calculation server CS for calculating eigenvalues connected to the communication node CN has been determined. At this stage, there is no local information notification NW determined for the communication NW1.

(Procedure S1-2-2) The NW administrator determines a maximum value Tmax of a desired required time from the occurrence of a failure / failure to the start of NW control. That is, Tmax is the maximum allowable value of the time for propagating the failure / failure information i.

(Procedure S1-2-3) In the calculation server CS, the maximum value Tmax of the required time is divided by the time TP2 necessary for communication between any two communication nodes belonging to the communication NW1 by the communication protocol P2. . Let Nmax be the maximum natural number that does not exceed the value obtained. Nmax is the maximum number of physical links that can transmit the failure / failure information i within the same time in the sense of (procedure S1-2) between the communication nodes on the local information notification NW using the protocol P2. Is shown.

(Procedure S1-2-4) An appropriate communication node N0 is determined from the communication NW1, and all the communication nodes up to Nmax that can be reached from the communication node N0 by following the physical link are collected.

(Procedure S1-2-5) The range constituted by the communication nodes collected in (Procedure S1-2-4) is defined as a range W1_n. At this time, the total number of communication nodes included in the range W1_n is N.

(Procedure S1-3) In the calculation server CS, an adjacency matrix A and a normalized Laplacian matrix NL in the range W1_n are generated. Both the adjacency matrix A and the normalized Laplacian matrix NL are N rows and N columns.

(Step S1-4) In the calculation server CS, the maximum eigenvalue λN of the adjacency matrix A and the second minimum eigenvalue λ2 of the normalized Laplacian matrix NL are obtained. Also, using [Equation 1], the initial arrival time π_W1_n when L = 2 and the order d (s) = d (t) = 1 is calculated.

  Next, the first arrival time π_W1_n is compared with Nmax, and whether or not π_W1_n <Nmax is satisfied, that is, the number of hops by random walk between any two nodes in the range W1_n is smaller than Nmax (contains within NW Make sure). Here, if π_W1_n ≧ Nmax, the number of communication nodes is increased one by one, and the processing after the procedure (1-2-4) is repeated so that π_W1_n <Nmax is satisfied.

(Procedure S1-5) If the maximum eigenvalue λN of the adjacency matrix A in the range W1_n and the second minimum eigenvalue λ2 of the normalized Laplacian matrix NL are increased in the calculation server CS (procedure S1-6) Proceed to On the other hand, when the maximum eigenvalue λN of the adjacency matrix A in the range W1_n and the second minimum eigenvalue λ2 of the normalized Laplacian matrix NL have not increased, the range W1_n is determined as the range W1_x. FIG. 5 is a diagram illustrating a state in which the range W1_x is determined.

(Procedure S1-6) In the calculation server CS, n = n + 1 is set, a communication node or / and a link other than the communication node CN belonging to W1_n is deleted, and a new range W1_n is configured. As a method for obtaining the new range W1_n, an optimization algorithm called an annealing method of Non-Patent Document 5 can be used.

(Procedure S1-7) When the range W1_x is determined in (Procedure S1-5), the calculation server CS uses the communication protocol CN to start the range W1_x for all nodes belonging to the range W1_x. The configuration information of the local information notification NW configured by is transmitted.

(Procedure S1-8) Perform the above (Procedure S1-2-1) to (Procedure S1-7) for the remaining NWs excluding the configured local information notification NW (range W1_x) of the communication NW1, Finally, the entire NW1 is divided by a local information notification NW such as W1_x (see FIG. 1).

  Thereafter, when failure / failure information i occurs in the local information notification NW, the failure / failure information i is sent to all nodes belonging to the range W1_x by the dedicated communication protocol P2 prepared in (Step S1-1). Processing to communicate is performed.

  Here, the transmission processing of the failure / failure information i by the communication node will be described in detail with reference to the drawings.

  FIG. 6 is a flowchart for explaining a local information propagation method (propagation stage) according to the first embodiment of this invention.

(Procedure S5-1) First, in each of the plurality of partial NWs configured in (Procedure S1-8), a shortest spanning tree having each node as a root is configured for all nodes in each partial NW. See Non-Patent Document 8 for the shortest spanning tree.

(Procedure S5-2) The communication node Nx belonging to the failure / failure range W1_x transmits the failure / failure information i to an adjacent node on the shortest spanning tree that is rooted by the flooding algorithm. See Non-Patent Document 6 for the flooding algorithm.

(Procedure S5-3) Other communication nodes that have received the failure / failure information i from the communication node Nx similarly transmit the failure / failure information i to an adjacent node on the shortest spanning tree that is rooted in itself. Each communication node repeats the above operation until failure / failure information i is transmitted to all communication nodes belonging to the range W1_x.

(Procedure S5-4) In (Procedure S5-3), one communication node may receive a plurality of pieces of failure / failure information i. In this case, the following (procedure S5-4-1) or (procedure S5-4-2) is performed.

(Procedure S5-4-1) A communication node that has received a plurality of pieces of failure / failure information i has different node identification information or occurrence time information included in the plurality of pieces of failure / failure information. The plurality of failure / failure information i is transmitted to an adjacent node.

(Procedure S5-4-2) On the other hand, when the node identification information and the occurrence time information included in the plurality of failure / failure information are the same, the communication node that has received the plurality of failure / failure information i One of the plurality of failure / failure information is selected and transmitted to the adjacent node on the shortest spanning tree.

  In this way, each communication node that receives the failure / failure information i performs control prescribed in advance by the NW administrator or the like.

  As described above, the two conflicting purposes of “setting the propagation area of local fault / failure information occurring in the NW to the widest and necessary part and propagating at high speed” can be achieved. This makes it possible to achieve the purpose of “performing local NW control at high speed”. In the above-described embodiment, an example of transmitting local failure / failure information that occurs in the NW has been described. However, local failure / failure information that occurs in various information processing systems and the like is also described. By the same procedure, it is possible to set a wide area and a necessary range and propagate it at high speed.

  Further, the route control protocol P1 described in (procedure S1-1) of the above-described embodiment may be an existing route control protocol such as a TCP / IP protocol (see Non-Patent Document 7).

  In addition to the annealing method, other optimization methods can be used as the optimization algorithm mentioned in (procedure S1-6) of the above-described embodiment. Furthermore, when using these optimization methods, the distribution of the degree, betweenness centrality, average distance, minimum loop size, and degree correlation of each node in the new range W_n (Non-Patent Document 5) is uniform. A communication node or / and link to be deleted may be selected.

[Second Embodiment]
Next, a second embodiment of the present invention in which congestion information j is transmitted instead of the failure / failure information i will be described in detail with reference to the drawings.

  FIG. 7 is a flowchart for explaining a local information propagation method (preparation stage) according to the second embodiment of this invention. Hereinafter, also in this embodiment, it is assumed that a communication network (hereinafter referred to as “communication NW2”) has already been configured, and the logical network NWL1 has been determined by path control. With respect to the local congestion information generated in the communication NW1, a procedure for setting a propagation area in a wide range and as much as possible and propagating at high speed will be described. In the following description, the congestion information includes the degree of congestion, identification information of the node where the congestion has occurred, the time of occurrence, and the like, and the communication node sets the local detour route based on the received congestion information. It shall have a control function.

(Procedure S2-1) When congestion occurs, a communication protocol P2 for transmitting congestion information j is prepared. As this communication protocol P2, a dedicated communication protocol capable of communicating at higher speed than the route control protocol P1 is selected. More preferably, a communication protocol suitable for spreading the congestion information j is selected using a route on an arbitrary part NW (local information notification NW) of the communication NW. For example, a protocol that spreads congestion information using the flooding algorithm of Non-Patent Document 6 may be used.

(Procedure S2-2) Next, assuming that n = 1, temporary congestion information notification for propagating the congestion information j of the communication NW as in (Procedure S2-2-1) to (Procedure S2-2-5) below The range W2_n for use is determined. Here, the range W2_n is a range in which the NW control function of all the communication nodes belonging to the range W2_n can recognize that the time when the congestion information j arrives is the same time. Hereinafter, in this range W2_n, W2_x satisfying the above (property 1) and (property 2) is obtained, and the local information notification NW is configured.

(Procedure S2-2-1) First, one calculation server CS for eigenvalue calculation connected to an arbitrary communication node CN not belonging to the configured local information notification NW of the communication NW is determined.

(Procedure S2-2-2) The NW administrator determines a maximum value Tmax of a desired required time from the occurrence of congestion to the start of NW control. That is, Tmax is the maximum allowable value for the time for which the congestion information j is propagated.

(Procedure S2-2-3) In the calculation server CS, the maximum value Tmax of the required time is divided by the time TP2 necessary for performing communication between any two communication nodes belonging to the communication NW2 by the communication protocol P2. . Let Nmax be the maximum natural number that does not exceed the value obtained. Nmax indicates the maximum number of physical links that can transmit the congestion information j within the same time in the meaning of (procedure S2-2) using the protocol P2 between communication nodes on the local information notification NW. ing.

(Procedure S2-2-4) An appropriate communication node N0 is determined from the communication NW2, and all the communication nodes up to Nmax that can be reached from this communication node N0 by following the physical link are collected.

(Procedure S2-2-5) The range constituted by the communication nodes collected in (procedure S2-2-4) is defined as a range W2_n. At this time, the total number of communication nodes included in the range W2_n is N.

(Procedure S2-3) In the calculation server CS, an adjacency matrix A and a normalized Laplacian matrix NL in the range W2_n are generated. Both the adjacency matrix A and the normalized Laplacian matrix NL are N rows and N columns.

(Step S2-4) In the calculation server CS, the maximum eigenvalue λN of the adjacency matrix A and the second minimum eigenvalue λ2 of the normalized Laplacian matrix NL are obtained. Also, the first arrival time π_W2_n when L = 2 and the order d (s) = d (t) = 1 is calculated using the above [Equation 1].

  Next, the first arrival time π_W2_n is compared with Nmax, and whether or not π_W2_n <Nmax, that is, the number of hops by random walk between any two nodes in the range W2_n is smaller than Nmax (contains within NW Make sure). In this case, if π_W2_n ≧ Nmax, the number of communication nodes is increased by one, and the processing after the procedure (2-2-4) is repeated so that π_W2_n <Nmax is satisfied.

(Procedure S2-5) In the calculation server CS, if the maximum eigenvalue λN of the adjacency matrix A in the range W2_n and the second minimum eigenvalue λ2 of the normalized Laplacian matrix NL appear to increase (procedure S2-6) Proceed to On the other hand, when the maximum eigenvalue λN of the adjacency matrix A in the range W2_n and the second minimum eigenvalue λ2 of the normalized Laplacian matrix NL have not increased, the range W2_n is determined as the range W2_x.

(Procedure S2-6) In the calculation server CS, n = n + 1 is set, a communication node or / and a link other than the communication node CN belonging to W2_n is deleted, and a new range W2_n is configured. As a method for obtaining the new range W2_n, the annealing method of Non-Patent Document 5 and other optimization algorithms can be used as in the first embodiment.

(Procedure S2-7) When the range W2_x is determined in (Procedure S2-5), the calculation server CS uses the communication protocol CN to start the range W2_x for all nodes belonging to the range W2_x. The configuration information of the local information notification NW configured by is transmitted.

(Procedure S2-8) Perform the above (Procedure S2-2-1) to (Procedure S2-7) for the remaining NWs excluding the configured local information notification NW (range W2_x) of the communication NW2, Finally, the entire NW2 is divided by a local information notification NW such as W2_x.

(S2-9 in FIG. 8) Thereafter, when congestion information j occurs in the local information notification NW, as shown in FIG. 8, the range is determined by the dedicated communication protocol P2 prepared in (procedure S2-1). A process of transmitting the congestion information j to all nodes belonging to W2_x is performed. Specific transmission processing of the congestion information j in each communication node is the same as that in the first embodiment (see FIG. 6).

(Procedure S2-10 in FIG. 8) In this way, each communication node that has received the congestion information j has a bypass route that is defined in advance by the NW administrator or the like so as not to pass through the communication node that caused the congestion. Then, normal communication is performed along the detour path.

  As described above, the present invention can be applied to the transmission of the congestion information j and the route detour control using the congestion information j.

  When the route control protocol P1 is an autonomous or autonomously distributed protocol, the route is not determined by the NW administrator or the like in advance as a bypass route in (procedure S2-10) in FIG. It may be constructed autonomously or autonomously distributed after arrival of the congestion information j by the control protocol P1.

[Third Embodiment]
Subsequently, instead of using a given NW, an NW that satisfies the above (property 1) and (property 2) (hereinafter referred to as “communication NW3”) is designed to transmit arbitrary local information k. A third embodiment of the present invention that has been made will be described in detail with reference to the drawings.

  FIG. 9 is a flowchart for explaining a local information propagation method (preparation stage) according to the third embodiment of this invention. The local information k includes the degree of the event, the identification information of the node where the event has occurred, the time of occurrence, and the like.

(Step S3-1) When a local event occurs, a communication protocol P2 for transmitting local information k is prepared. As this communication protocol P2, a dedicated communication protocol capable of communicating at higher speed than the route control protocol P1 is selected. More preferably, a communication protocol suitable for spreading local information k is selected using a route on an arbitrary part NW (local information notification NW) of communication NW3. For example, a protocol that spreads the local information k by the flooding algorithm of Non-Patent Document 6 may be used.

(Procedure S3-2) Next, as n = 1, temporary local information notification for propagating the local information k of the communication NW3 as in (Procedure S3-2-1) to (Procedure S3-2-6) A range W3_n is determined. Here, the range W3_n is a range in which the NW control function possessed by all communication nodes belonging to the range W3_n can recognize the time when the local information k arrives as the same time. Hereinafter, in this range W3_n, W3_x that satisfies the above (property 1) and (property 2) is obtained, and the local information notification NW is configured.

(Procedure S3-2-1) First, the NW designer sets the maximum value n_max of the number of nodes of the communication NW3 to be designed and the maximum value L_max of the number of links in the NW design server.

(Procedure S3-2-2) Next, the NW designer determines a maximum value Tmax of a desired required time from the local information generation time point until the NW control is started. That is, Tmax is the maximum allowable value for the time for propagating the local information k.

(Procedure S3-2-3) The maximum value Tmax of the required time is divided by the time TP2 necessary for performing communication between any two communication nodes belonging to the communication NW3 by the communication protocol P2. Let Nmax be the maximum natural number that does not exceed the value obtained. Nmax indicates the maximum number of physical links capable of transmitting local information k within the same time in the sense of (procedure S3-2) between the communication nodes on the local information notification NW using protocol P2. ing.

(Procedure S3-2-4) The NW design server configures a complete graph NWC with the number of nodes n_max.

(Procedure S3-2-5) The NW design server determines an appropriate communication node N0 from the communication NW3, and collects all Nmax communication nodes that can be reached from the communication node N0 by following the physical link.

(Procedure S3-2-6) The range constituted by the communication nodes collected in (procedure S3-2-5) is defined as a range W3_n. At this time, the total number of communication nodes included in the range W3_n is N.

(Procedure S3-3) In the NW design server, an adjacency matrix A and a normalized Laplacian matrix NL in the range W3_n are generated. Both the adjacency matrix A and the normalized Laplacian matrix NL are N rows and N columns.

(Step S3-4) In the NW design server, the maximum eigenvalue λN of the adjacency matrix A and the second minimum eigenvalue λ2 of the normalized Laplacian matrix NL are obtained. Also, the first arrival time π_W3_n when L = 2 and the order d (s) = d (t) = 1 is calculated using the above [Equation 1].

  Next, the NW design server compares the first arrival times π_W3_n and Nmax, and whether or not π_W3_n <Nmax, that is, the number of hops by random walk between any two nodes in the range W3_n is smaller than Nmax ( Confirm that it is within the NW. Here, if π_W3_n ≧ Nmax, the number of communication nodes is increased one by one, and the processing after the procedure (3-2-4) is repeated so that π_W3_n <Nmax is satisfied.

(Procedure S3-5) If the maximum eigenvalue λN of the adjacency matrix A in the range W3_n and the second minimum eigenvalue λ2 of the normalized Laplacian matrix NL appear to increase (procedure S3-5) (procedure S3-6) Proceed to On the other hand, when the maximum eigenvalue λN of the adjacency matrix A in the range W3_n and the second minimum eigenvalue λ2 of the normalized Laplacian matrix NL have not increased, the number of nodes and the number of links in the range W3_n are respectively (step S3-2- It is confirmed whether it is below n_max and L_max determined in 1).

  As a result of the confirmation, if the number of nodes and the number of links in the range W3_n seem to satisfy the conditions defined in (Step S3-2-1), the NW design server determines the range W3_n as the range W3_x. . If at least one of the number of nodes and the number of links in the range W3_n does not satisfy the condition defined in (Procedure S3-2-1), the NW design server proceeds to (Procedure S3-6).

(Procedure S3-6) In the NW design server, n = n + 1 is set, a communication node or / and link other than the communication node CN belonging to W3_n is deleted, and a new range W3_n is configured. As a method for obtaining the new range W3_n, the annealing method and other optimization algorithms of Non-Patent Document 5 can be used as in the first and second embodiments.

(Procedure S3-7) When the range W3_x is determined in (Procedure S3-5), the NW design server uses the communication protocol CN to start the range W3_x for all nodes belonging to the range W3_x. The configuration information of the local information notification NW configured by is transmitted and stored.

(Procedure S3-8) Perform the above (Procedure S3-2-4) to (Procedure S3-7) for the remaining NWs excluding the configured local information notification NW (range W3_x) of the communication NW3, Finally, the communication NW3 divided by the local information notification NW, such as the whole W3_x, is constructed.

(S3-9 in FIG. 10) When the local information k is generated in the NW3 local information notification NW configured as described above, the local information k is prepared in (procedure S3-1) as shown in FIG. A process for transmitting the local information k to all nodes belonging to the range W3_x is performed by the dedicated communication protocol P2. The specific transmission process of the local information k in each communication node is the same as in the first embodiment (see FIG. 6).

(Procedure S3-10 in FIG. 10) In this way, each communication node that receives the local information k performs NW control determined in advance by an NW administrator or the like.

  As described above, by designing the network so that the local information transmission method of the present invention can be suitably implemented, the subsequent determination of the propagation range W3 of the local information k can be omitted.

  When each communication node has an autonomous or autonomous distributed NW control function, the NW control in FIG. 10 (procedure S3-10) is determined in advance by the NW administrator or the like. Instead, the control content corresponding to the local information k may be executed by the NW control function.

[Fourth Embodiment]
Subsequently, an NW (hereinafter referred to as “communication NW4”), which has a function of providing a function F virtually to another node in the NW and capable of transmitting the request information m of the function F, The fourth embodiment of the present invention that constitutes the design of the above will be described in detail with reference to the drawings.

  FIG. 11 is a flowchart for explaining a request information propagation method (preparation stage) according to the fourth embodiment of this invention. The request information m includes the requested function, the requesting node identification information, the request time, and the like.

(Procedure S4-1) When the request information m for the function F occurs, a communication protocol P2 for transmitting the request information m is prepared. As this communication protocol P2, a dedicated communication protocol capable of communicating at higher speed than the route control protocol P1 is selected. More preferably, a communication protocol suitable for spreading the request information m is selected using a route on an arbitrary part NW (local information notification NW) of the communication NW4. For example, a protocol that spreads the request information m by the flooding algorithm of Non-Patent Document 6 may be used.

(Procedure S4-2) Next, assuming that n = 1, the range for temporary local information notification in which the request information m is propagated as in (Procedure S4-2-1) to (Procedure S4-2-6) below W4_n is determined. Here, the range W4_n is a range in which the NW control function possessed by all the communication nodes belonging to the range W4_n can recognize the time when the request information m arrives as the same time. Hereinafter, in this range W4_n, W4_x satisfying the above (property 1) and (property 2) is obtained, and the local information notification NW is configured.

(Procedure S4-2-1) First, the NW designer sets the maximum value n_max of the number of nodes of the communication NW4 to be designed and the maximum value L_max of the number of links in the NW design server.

(Procedure S4-2-2) Next, the NW designer determines a maximum value Tmax of a desired required time from the local information generation time point until the NW control is started. That is, Tmax is the maximum allowable value for the time for propagating the request information m.

(Procedure S4-2-3) The maximum value Tmax of the required time is divided by the time TP2 required for performing communication between any two communication nodes belonging to the communication NW4 by the communication protocol P2. Let Nmax be the maximum natural number that does not exceed the value obtained. Nmax indicates the maximum number of physical links that can transmit the request information m within the same time in the sense of (procedure S4-2) using the protocol P2 between the communication nodes on the local information notification NW. ing.

(Procedure S4-2-4) The NW design server configures a complete graph NWC with the number of nodes n_max.

(Procedure S4-2-5) The NW design server determines an appropriate communication node N0 from the communication NW4, and collects all Nmax communication nodes that can be reached from the communication node N0 by following the physical link.

(Procedure S4-2-6) The range constituted by the communication nodes collected in (procedure S4-2-5) is defined as a range W4_n. At this time, the total number of communication nodes included in the range W4_n is N.

(Procedure S4-3) In the NW design server, an adjacency matrix A and a normalized Laplacian matrix NL in the range W4_n are generated. Both the adjacency matrix A and the normalized Laplacian matrix NL are N rows and N columns.

(Step S4-4) In the NW design server, the maximum eigenvalue λN of the adjacency matrix A and the second minimum eigenvalue λ2 of the normalized Laplacian matrix NL are obtained. Also, the first arrival time π_W4_n when L = 2 and the order d (s) = d (t) = 1 is calculated using the above [Equation 1].

  Next, the NW design server compares the first arrival times π_W4_n and Nmax, and whether or not π_W4_n <Nmax, that is, the number of hops by random walk between any two nodes in the range W4_n is smaller than Nmax ( Confirm that it is within the NW. In this case, if π_W4_n ≧ Nmax, the number of communication nodes is increased by one, and the processing after the procedure (4-2-4) is repeated so that π_W4_n <Nmax is satisfied.

(Procedure S4-5) If the maximum eigenvalue λN of the adjacency matrix A in the range W4_n and the second minimum eigenvalue λ2 of the normalized Laplacian matrix NL appear to increase (procedure S4-5) (procedure S4-6) Proceed to On the other hand, when the maximum eigenvalue λN of the adjacency matrix A in the range W4_n and the second minimum eigenvalue λ2 of the normalized Laplacian matrix NL are not increased, the number of nodes and the number of links in the range W4_n are respectively (step S4-2-2). It is confirmed whether it is below n_max and L_max determined in 1).

  As a result of the confirmation, if the number of nodes and the number of links in the range W4_n seem to satisfy the conditions defined in (Step S4-2-1), the NW design server determines the range W4_n as the range W4_x. . If at least one of the number of nodes and the number of links in the range W4_n does not satisfy the condition defined in (Procedure S4-2-1), the NW design server proceeds to (Procedure S4-6).

(Procedure S4-6) In the NW design server, n = n + 1 is set, a communication node or / and a link other than the communication node CN belonging to W4_n is deleted, and a new range W4_n is configured. As a method for obtaining the new range W4_n, the annealing method and other optimization algorithms described in Non-Patent Document 5 can be used as in the first to third embodiments.

(Procedure S4-7) When the range W4_x is determined in (Procedure S4-5), the NW design server uses the communication protocol CN as a starting point for the range W4_x for all nodes belonging to the range W4_x. The configuration information of the local information notification NW configured by is transmitted and stored.

(Procedure S4-8) Perform the above (Procedure S4-2-4) to (Procedure S4-7) for the remaining NWs excluding the configured local information notification NW (range W4_x) of the communication NW4, Eventually, NW4 divided by the local information notification NW, such as W4_x as a whole, is constructed.

(S4-9 in FIG. 12) When the request information m is generated in the local information notification NW of NW4 configured in this way, as shown in FIG. 12, it is prepared in (procedure S4-1). A process for transmitting the request information m to all nodes belonging to the range W4_x is performed by the dedicated communication protocol P2. The specific transmission process of the request information m in each communication node is the same as that in the first embodiment (see FIG. 6).

(Procedure S4-10 in FIG. 12) In this way, if the communication node that has received the request information m has the function F, the function F is transmitted to the node that has transmitted the request information m on the NW. Provide virtually.

  As described above, by designing the network so that request information can be suitably transmitted to a node that virtually provides a function via the NW, the subsequent propagation range W4 of the request information m can be determined. Omitted, the function F can be provided at high speed.

  The preferred embodiments of the present invention have been described above. However, the present invention is not limited to the above-described embodiments, and further modifications, replacements, and replacements may be made without departing from the basic technical idea of the present invention. Adjustments can be made. For example, in the above-described embodiment, the example used for local information transmission between NW nodes has been described. However, the same applies to local information generated in various information processing systems. According to the procedure, a wide area and a necessary range can be set and can be propagated at high speed.

  Each disclosure of the above non-patent document is incorporated herein by reference. Within the scope of the entire disclosure (including claims) of the present invention, the embodiment can be changed and adjusted based on the basic technical concept. Further, various combinations or selections of various disclosed elements (including each element of each claim, each element of each embodiment, each element of each drawing, etc.) are possible within the scope of the claims of the present invention. is there. That is, the present invention of course includes various variations and modifications that could be made by those skilled in the art according to the entire disclosure including the claims and the technical idea.

11 Communication unit 12 Laplacian matrix calculation unit 13 Second minimum eigenvalue calculation unit 14 Adjacency matrix calculation unit 15 Maximum eigenvalue calculation unit 16 First arrival time calculation unit 17 Comparison unit 18 NW configuration determination unit A Adjacency matrix NL Normalized Laplacian matrix λN Adjacency matrix The maximum eigenvalue λ2 The second minimum eigenvalue NWC of the normalized Laplacian matrix Complete graph P1 with n_max number of nodes Protocol P2 for route control Dedicated protocol CN for propagating local information Communication node CS Calculation server i Fault / failure information j Congestion Information k Local information m Request information

Claims (10)

Maximizing the maximum eigenvalue of the adjacency matrix representing the partial network so that the partial network propagating local information generated in the network has a predetermined size, and the second minimum eigenvalue of the normalized Laplacian matrix of the partial network Calculating a partial network configuration that maximizes
Notifying all the nodes in the partial network of the calculated configuration information;
Propagating the local information to the partial network using the configuration information when the local information occurs.   The local information propagation method according to claim 1, wherein a dedicated protocol defined for the propagation of the local information is used for the propagation of the local information.   The upper limit of the size of the partial network is obtained by dividing the maximum allowable time for propagating the local information by the time required for communication between any two communication nodes using the protocol defined for the propagation of the local information. The method for propagating local information according to claim 1 or 2 obtained by The local information includes the degree of the event, the identification information of the node where the event has occurred, the occurrence time,
The local information propagation method according to claim 1, further comprising a step in which a node of the network performs control of the network according to the local information. The local information includes the degree of congestion, the identification information of the node where the congestion has occurred, and the time of occurrence,
The local information propagation method according to claim 1, further comprising a step in which a node of the network performs path control according to the local information. Instead of the partial network that propagates local information generated in the network, calculate the configuration of the partial system that propagates local information generated in the system,
Notifying all the nodes in the partial system of the calculated configuration information;
The local information propagation method according to claim 1, wherein, when local information is generated, the local information is propagated to the partial system using the configuration information. As a partial network for propagating local information generated in the network, a partial network that maximizes the maximum eigenvalue of the adjacency matrix representing the network and maximizes the second minimum eigenvalue of the normalized Laplacian matrix of the network Means for calculating the configuration of
Means for notifying all the nodes in the partial network of the calculated configuration information, and a calculation node comprising:
A network in which, when local information is generated, a node belonging to the partial network propagates the local information using a protocol defined for the propagation of the local information. As a partial network for propagating local information generated in the network, a partial network that maximizes the maximum eigenvalue of the adjacency matrix representing the network and maximizes the second minimum eigenvalue of the normalized Laplacian matrix of the network Means for calculating the configuration of
Means for notifying all the nodes in the partial network of the calculated configuration information.   A node that propagates the local information using a predetermined protocol by using the configuration information communicated from the calculation node according to claim 8 when local information is generated. Maximizing the maximum eigenvalue of the adjacency matrix representing the partial network so that the partial network propagating local information generated in the network has a predetermined size, and the second minimum eigenvalue of the normalized Laplacian matrix of the partial network A process of calculating a partial network configuration that maximizes
A program for causing a calculation node connected to the network to execute a process of notifying the calculated configuration information to all nodes in the partial network.

Images