WO2023007552A1

WO2023007552A1 - Communication network model parameter setting assistance device, communication network model parameter setting assistance method, and program

Info

Publication number: WO2023007552A1
Application number: PCT/JP2021/027570
Authority: WO
Inventors: 智洋郡川; 雅史清水; 直樹高谷; 英成大和田; 恭太服部
Original assignee: 日本電信電話株式会社
Priority date: 2021-07-26
Filing date: 2021-07-26
Publication date: 2023-02-02
Also published as: JPWO2023007552A1

Abstract

In a communication network model parameter setting assistance device (10), node models (21a), corresponding to actual nodes (32a) of different node types in an actual network (31) in which the actual nodes (32a) are deployed through network connection, are connected to each other. Node model parameters corresponding to parameters of the actual nodes 32a are set to the node models 21a. The performances of single bodies of the actual nodes 32a and the node models 21a operated by input of a traffic pattern are measured, and performance measurement information items, node internal information items, and statistic information items are collected. In accordance with the information items of the measurement and the collection, a node model parameter corresponding to an assessment point at which the absolute value of a difference between assessment results of the single bodies becomes smallest is inferred as the closest parameter which is closest to an actual node (32a).

Description

Communication network model parameter setting support device, communication network model parameter setting support method and program

According to the present invention, for each node model corresponding to each real node constituting a communication network, the parameters closest to the internal configuration such as the number of cores and clocks of the processor of each real node, the number of various queues and the queue length are set to each real node model. The present invention relates to a communication network model parameter setting support device, a communication network model parameter setting support method, and a program that support setting in each node model of a network model corresponding to a node.

An actual communication network is constructed by connecting nodes (also referred to as real nodes) comprising communication devices such as servers, routers, and switches via a wired or wireless network {also referred to as NW (Network)}. A real communication network is also called a real NW.

Conventionally, there is a technology (Non-Patent Documents 1 and 2) that configures the NW topology, which is the connection relationship between NW nodes, from the setting information of each node that constitutes the actual communication NW. There is also a technique of estimating the setting parameters of a single node model of the NW model by assuming a mathematical model such as a queue and by measuring the performance of real nodes of the real NW (Non-Patent Documents 3 and 4).

By the way, in order to bring the accuracy of the NW model closer to the real NW, parameters that are closer to the internal configuration such as the number of cores of the processor that the real node of the real NW has, the number of clocks, the number of various queues, the queue length, etc. Must be set. However, the techniques of Non-Patent Documents 1 and 2 are limited to modeling the NW topology, which is the connection relationship between real nodes.

Also, it is necessary to set parameters for the node model so as to correspond to multiple types of real nodes that make up the real NW. However, the techniques of Non-Patent Documents 3 and 4 are limited to parameter estimation related to a specific modeling method corresponding to a single type of node.

For these reasons, there was the problem that it was not possible to set parameters that approximate the real nodes that make up the real NW in the node models that make up the NW model.

The present invention has been made in view of such circumstances, and an object of the present invention is to set parameters that approximate the real nodes that make up the real NW in the node models that make up the NW model.

In order to solve the above problems, a communication network model parameter setting support device of the present invention corresponds to a real NW in which real nodes of different node types as communication devices are connected via a NW (Network) and deployed. A network model to which a corresponding node model is connected, a setting unit that sets parameters for the node model corresponding to the real node to the node model, and a traffic pattern that changes the node load, the node model and the real node. a load test unit for inputting data to a load test unit, a performance measurement unit for measuring the performance of each of the node model unit and the real node unit that operate on the input of the traffic pattern to obtain performance measurement information for each, the node model unit and a collection unit for collecting the node internal information and statistical information of each of the real nodes, and the performance measurement information and the node internal information and statistical information of the individual node models and the individual real nodes in an evaluation space in one node type; Corresponds to the evaluation point with the minimum absolute value of the difference between the evaluation results of the node model alone and the actual node alone recorded in the confirmation part and the recording part as the node evaluation results related to the above evaluation points. an analysis unit for estimating the parameter for the node model to be the nearest parameter closest to the real node.

According to the present invention, it is possible to set parameters that approximate the real nodes that make up the real NW in the node models that make up the NW model.

1 is a block diagram showing the configuration of a communication network model parameter setting support device according to a first embodiment of the present invention; FIG. 4 is a flowchart for explaining the operation of communication NW modeling support processing by the communication network model parameter setting support device according to the first embodiment; FIG. 9 is a block diagram showing the configuration of a communication network model parameter setting support device according to the second embodiment of the present invention; 10 is a flowchart for explaining the operation of communication NW modeling support processing by the communication network model parameter setting support device according to the second embodiment;

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described with reference to the drawings.
<Configuration of First Embodiment>
FIG. 1 is a block diagram showing the configuration of a communication network model parameter setting support device according to the first embodiment of the present invention.

A communication network model parameter setting support device (also referred to as a support device) 10 shown in FIG. Perform parameter setting support processing. The real NW 31 is an actual communication network. The NW model 21 is constructed on the NW simulator.

However, the real NW 31 is constructed by connecting the real node 32a to the network. The real node 32a is a communication device such as a server, router, or switch. The NW model 21 is configured by virtually network-connecting a node model 21a corresponding to the real node 32a on a network simulator (not shown).

The support device 10 includes a node information collection unit 20, a NW model 21, a load test unit 22, a parameter setting unit 23, a performance measurement unit 24, an evaluation control unit 25, an evaluation space information DB unit 26, an evaluation It comprises a result confirmation section 27 , an evaluation result recording section 28 , an evaluation result analysis section 29 and a parameter policy section 30 .

An evaluation space information DB unit (also referred to as a DB unit) 26 stores types and values of parameters (to be described later) for a node model corresponding to the real node 32a, traffic patterns (traffic pattern information) to be described later, and the like. ing. In other words, the DB unit 26 stores evaluation space information including parameters and traffic patterns.

The above parameters include, for example, the number of cores of the processor, the number of clocks, cache capacity, etc. as parameter types. Further enumerating types of parameters include memory capacity, number of channels, types {SRAM (Static Random Access Memory), DRAM (Dynamic Random Access Memory), CAM (Content Addressable Memory), etc.}. There is also backplane bandwidth (the bandwidth over which the router can operate). There is also a number of internal table entries (routes, filters, etc.). Also, if the node is mobile, there is battery capacity. There are number of ports and number of links. There are types and versions of OS (Operation System). Furthermore, if the node is a router, there are the number of ports used by the router, the number of queues inside the router, and the length of each queue.

There are values (parameter values) such as "1, 2, 4, 8, 16, 32" as the types of these parameters, for example, the number of processor cores.

Information that characterizes the above traffic patterns includes frame size, packet size, rate, layer, protocol, presence or absence of bursts, number of flows, and so on. These traffic patterns also have values (traffic pattern values) such as frame sizes of "64, 128, 512, 768, 1024, 1518" bytes. For that frame size, the more values, the higher the measurement quality, but the longer it takes.

This traffic pattern is a pattern such as an input packet whose node load changes according to the input of the traffic pattern. For example, when a large number of short packets are input at predetermined time intervals and when a small number of long packets are input, the load on the router differs even if the total traffic volume is the same. In this way, differences in the amount and size of input packets, the number of input packets per unit time, the protocol followed by the packets, and the flow information such as the destination address included in the input packets cause differences in the load on the nodes.

We will explain the evaluation space in the evaluation space information as such parameters and traffic patterns. The evaluation space is a space corresponding to a mathematical set defining evaluation points j=1 to Li (maximum Li) for evaluating nodes. This evaluation space is defined for each type of node (node type). It is assumed that the number of node types i is i=1 to M. In other words, in each of the i=1 to M-th node types, the j=1 to Li-th evaluation points j on the evaluation space are evaluated.

The evaluation control unit 25 reads parameters and traffic patterns corresponding to the evaluation point j on the evaluation space for the node type i from the DB unit 26 . The read parameters are output to a parameter setting section (also referred to as a setting section) 23 and set in the node model 21 a of the NW model 21 by the setting section 23 . Also, the traffic pattern read by the evaluation control section 25 is output to the load test section 22 .

The load test unit 22 inputs the input traffic pattern to the node model 21 a of the NW model 21 and the real node 32 a of the real NW 31 .

The performance measurement unit 24 measures performance such as throughput, delay, and packet discard rate during operation of the node model 21a and the real node 32a to which the traffic pattern is input. The result of measuring the performance of the node model 21a alone is referred to as model performance measurement information, and the result of measuring the performance of the real node 32a alone is referred to as actual performance measurement information. The model performance measurement information and the actual performance measurement information are output to the evaluation result confirming section 27 .

The node information collection unit 20 collects setting information of the node model 21a and the real node 32a, log information, statistics inside the node such as the number of packet arrivals and discards in each queue inside the node, processor usage rate, memory usage rate, etc. collect information and These collected node internal information and statistical information are output to the evaluation result confirming unit 27 .

An evaluation result confirming unit (also referred to as a confirming unit) 27 determines an evaluation point j (for example, j= Second), the evaluation result of the node model 21a and the evaluation result of the real node 32a are associated with each other and output to an evaluation result recording unit (also referred to as a recording unit) . The recording unit 28 associates and records the evaluation result of the node model 21a and the evaluation result of the actual node 32a regarding the evaluation point j (eg, j=2nd) in the evaluation space in the node type i (eg, i=1st). do.

Further, when the model performance measurement information and the actual performance measurement information are input from the performance measurement unit 24, the confirmation unit 27 checks the evaluation point j (for example, j=1) in the evaluation space for the node type i (for example, i=1st 2) is notified to the evaluation control unit 25. At the time of this notification, the evaluation of the second evaluation point on the evaluation space for the first node type ends.

When the evaluation control unit 25 receives notification from the confirmation unit 27 that the evaluation of all evaluation points j=Li on the evaluation space related to one node type i (for example, i=1st) has been completed, the next node Parameters and traffic patterns for evaluating evaluation points j=1 to Li-th on the evaluation space for species i=2 are read from the DB unit 26 . Also, the evaluation control unit 25 terminates the evaluation process when receiving notification from the confirmation unit 27 that evaluation of all evaluation points j=Li on the evaluation space related to the last node type i=Mth has been completed.

When the evaluation control unit 25 notifies the evaluation result analysis unit 29 of the completion of the evaluation of all the evaluation points j=Li on the evaluation space for one node type, the evaluation result analysis unit 29 stores the node model 21a recorded in the recording unit 28 and the actual node model 21a. Analysis for estimating the latest parameters described above is performed as follows based on each evaluation result of evaluation points j=1 to Li of both nodes 32a.

That is, the analysis unit 29 searches for the evaluation point j on the evaluation space that minimizes the absolute value of the difference between the evaluation result of the real node 32a and the evaluation result of the node model 21a, and calculates the parameter corresponding to the evaluation point j. Let it be the most recent parameter. This recent parameter is output to the parameter policy section 30 .

However, in the latest parameter estimation by the analysis unit 29, when the absolute value of the difference between the evaluation result of the node model 21a and the evaluation result of the real node 32a becomes equal to or less than a predetermined threshold, the evaluation in the entire evaluation space The parameters at this point in time may be estimated as recent parameters without waiting for the completion of the evaluation for point k. If the absolute value of the difference between the evaluation result of the node model 21a and the evaluation result of the real node 32a does not become equal to or less than the threshold even after the evaluation for the evaluation point k on the entire evaluation space is completed, a new evaluation on the entire evaluation space is performed. Additional points may be set to expand the total evaluation space.

A parameter policy unit (also referred to as a policy unit) 30 stores recent parameters from the analysis unit 29. The stored recent parameters are one node type when searching for recent parameters for each node type i=1 to M. be the initial parameter for i. The initial parameters are output to the evaluation control unit 25 and set in the node model 21a by the setting unit 23. FIG.

<Operation of the first embodiment>
Next, the first communication network model parameter setting support processing by the support device 10 of the first embodiment will be described with reference to the flowchart shown in FIG.

However, although one real node 32a in the real NW 31 and one node model 21a in the NW model 21 are shown in FIG. shall exist. It is assumed that the evaluation space information DB unit 26 stores evaluation space information including traffic patterns and M types of node model parameters corresponding to M types of real node 32a parameters.

In step S1 shown in FIG. 2, the evaluation control unit 25 initializes i=1, and in step S2, the evaluation control unit 25 initializes j=1. However, i corresponds to the node type of the node model 21a and the real node 32a, and is varied from i=1 to M one by one. j corresponds to an evaluation point on the evaluation space and is varied one by one from j=1 to Li.

In step S3, the evaluation control unit 25 reads from the DB unit 26 parameters and traffic patterns corresponding to the evaluation point j=1 on the evaluation space for the node type i=1. The read parameters are output to the parameter setting unit 23 and set in the node model 21 a of the NW model 21 by the setting unit 23 . Also, the traffic pattern read by the evaluation control section 25 is output to the load test section 22 .

In step S4, the load test unit 22 inputs the input traffic pattern to the node model 21a and the real node 32a for a predetermined period of time.

In step S5, the performance measurement unit 24 measures performance such as throughput, delay, and packet discard rate during operation of the node model 21a and the real node 32a to which the traffic pattern is input. The measured model performance measurement information and actual performance measurement information are output to the confirmation unit 27 .

In step S6, the collection unit 20 collects the setting information of the node model 21a and the real node 32a, log information, the number of packet arrivals and discards in each queue inside the node, the usage rate of the processor, the usage rate of the memory, etc. to collect statistics and The collected node internal information and statistical information are output to the confirmation unit 27 .

In step S7, the confirmation unit 27 converts the model performance measurement information, the node internal information, and the statistical information of the node model 21a to the evaluation result of the node related to the evaluation point j=1 in the evaluation space for the node type i=1. get. At the same time, the confirmation unit 27 acquires the actual performance measurement information, the node internal information, and the statistical information of the real node 32a as the evaluation result of the node related to the evaluation point j=1st in the evaluation space for the node type i=1st. . The confirmation unit 27 associates the evaluation results of both the node model 21a and the real node 32a with respect to the evaluation point j=1 on the evaluation space in the node type i=1 and outputs them to the recording unit 28 . Further, when the model performance measurement information, the actual performance measurement information, the node internal information, and the statistical information are input from the performance measurement unit 24, the confirmation unit 27 confirms that the node type i=1-th evaluation point j= on the evaluation space The evaluation control unit 25 is notified of the end of the first evaluation. At the time of this notification, the evaluation of the first evaluation point on the evaluation space for the first node type ends.

In step S8, upon receiving the notification from the confirmation unit 27, the evaluation control unit 25 determines whether or not j=Li. In this case, it is not j=Li (No), so returning to step S3, the evaluation control unit 25 obtains the evaluation point j=j+1 (here, j= Read the parameters and traffic patterns corresponding to 2). After that, the processing of steps S4 to S8 is repeated.

After that, in step S8, the evaluation control unit 25 determines whether or not j=Li, and if the result is j=Li (Yes), the process proceeds to step S9.

In step S9, when the analysis unit 29 is notified from the evaluation control unit 25 that the evaluation of all the evaluation points j=Li on the evaluation space for one node type has been completed, the analysis unit 29 =1 to Li, the analysis for estimating the latest parameters, which will be described later, is performed as follows.

That is, for each evaluation result, the analysis unit 29 searches for the evaluation point j on the evaluation space that minimizes the absolute value of the difference between the evaluation results of both the real node 32a and the node model 21a, and sets the evaluation point j to Let the corresponding parameter be the latest parameter. This recent parameter is output to the parameter policy unit 30 and stored. The saved recent parameters are output to the evaluation control unit 25 as initial parameters for one node type i when searching for recent parameters for each node type i=1 to M. FIG.

In step S10, when the evaluation control unit 25 receives the latest parameter of the node type i=1, it sets the latest parameter to the corresponding node model 21a via the setting unit 23.

At step S10A, the evaluation control unit 25 determines whether or not the node type i=M. In this case, since i is not equal to M (No), it is set to i=i+1 (here, i=2), and after returning to step S2 and executing the processing, steps S3 to S10 are repeated.

After that, in step S10A, the evaluation control unit 25 determines whether or not i=M, and if the result is i=M (Yes), the first communication NW model parameter setting support processing ends.

<Effects of the first embodiment>
Effects of the support device 10 according to the first embodiment of the present invention will be described.

The support device 10 includes a NW model 21, a setting unit 23, a load test unit 22, a performance measurement unit 24, a collection unit 20, a confirmation unit 27, a recording unit 28, and an analysis unit 29.

The NW model 21 corresponds to the real NW 31 in which real nodes 32a of different node types as communication devices are network-connected and deployed, and the node models 21a corresponding to the real nodes 32a are connected.

The setting unit 23 sets parameters for the node model corresponding to the parameters of the real node 32a in the node model 21a. The load testing unit 22 inputs a traffic pattern that changes the node load to the node model 21a and the real node 32a. The performance measuring unit 24 measures the performance of each of the node model 21a alone and the real node 32a alone, which are operated by the input of the traffic pattern, and obtains performance measurement information of each. The collection unit 20 collects node internal information and statistical information of each of the node model 21a and the real node 32a.

The confirmation unit 27 acquires the performance measurement information, the node internal information, and the statistical information of each of the node model 21a single unit and the real node 32a single unit as the node evaluation results related to the evaluation points in the evaluation space for one node type, and records the 28.

The analysis unit 29 finds the minimum absolute value of the difference between the evaluation results of the single node 21a and the single real node 32a among the node evaluation results related to all the evaluation points on the evaluation space for one node type recorded in the recording unit 28. The parameter for the node model corresponding to the evaluation point is estimated as the parameter (most recent parameter) closest to the configuration of the real node 32a.

According to this configuration, the parameters (most recent parameters) that are closest to the configuration of the real nodes 32a that make up the real NW 31 can be estimated and set in the node models 21a that make up the NW model 21.

<Configuration of Second Embodiment>
FIG. 3 is a block diagram showing the configuration of a communication network model parameter setting support device according to the second embodiment of the present invention.

The support device 10A of the second embodiment shown in FIG. 3 differs from the support device 10 (FIG. 1) of the first embodiment described above in that all node models (all node models) 21a to 21a of one NW model 21A 21d, the parameters are estimated recently.

The support device 10A includes a node information collection unit 20A, a NW model 21A, a load test unit 22A, a parameter setting unit 23A, a performance measurement unit 24A, an evaluation control unit 25A, an evaluation space information DB unit 26A, an evaluation It comprises a result confirmation section 27A, an evaluation result recording section 28A, an evaluation result analysis section 29A, and a parameter policy section 30A.

However, the real NW 31A is constructed by network-connecting real nodes 32a to 32d of different node types (plurality). It is also assumed that the aforementioned evaluation space has evaluation points k=1 to L (maximum L) for evaluating the NW model and the actual NW.

The NW model 21A is configured by virtually network-connecting a plurality of node models 21a to 21d corresponding to a plurality of real nodes 32a to 32d on a network simulator (not shown).

The DB section 26A stores the parameters for the node model corresponding to the parameters of the real nodes 32a to 32d and the evaluation space information including the traffic pattern.

The policy unit 30A stores the latest parameter of each node type i as an initial parameter in the same manner as the policy unit 30 (FIG. 1) described above, and stores the priority in searching for the parameter in association with each parameter. . In the policy unit 30A, when estimating the latest parameters for all the node models 21a to 21d, it takes time to search for the evaluation point k=L in the entire evaluation space. has a higher priority.

　Parameter sensitivity (parameter sensitivity) means that, for example, if a node has no battery, entering the remaining battery power as a parameter does not affect node processing. Therefore, the remaining battery parameter is less sensitive. On the other hand, since the number of clocks has a great influence on the processing of the node, the sensitivity is high. By giving high priority to such parameters with high parameter sensitivities, it is possible to shorten the estimation time when estimating the latest parameters in all the node models 21a to 21d.

In addition, the policy unit 30A can select the next evaluation point from unsearched evaluation points in the evaluation space search history based on the priority of parameter search. For example, the analysis unit 29A evaluates all the evaluation spaces (the number of evaluation spaces for each node type×node types), and evaluates all the node models 21a to 21d in order from the points including the most sensitive parameter of the node type. to implement.

The evaluation control unit 25A reads the initial parameters corresponding to the evaluation points k on the evaluation space for all the node models 21a to 21d from the policy unit 30A, and also reads the traffic patterns from the DB unit 26A. The read parameters are output to the setting section 23A, and are set in the respective node models 21a to 21d of the NW model 21A by the setting section 23A. Also, the traffic pattern read by the evaluation control section 25A is output to the load testing section 22A.

The load test unit 22A inputs the input traffic pattern to the NW model 21A and the actual NW 31A.

The performance measurement unit 24A measures performance such as throughput, delay, and packet discard rate during operation between the NW model 21A to which the traffic pattern is input and the real NW 31A. The result of measuring the performance of the NW model 21A is called NW model performance measurement information, and the result of measuring the performance of the real NW 31A is called real NW performance measurement information. The NW model performance measurement information and the actual NW performance measurement information are output to the confirmation section 27A.

The node information collection unit 20A collects node internal information and statistical information of all node models 21a to 21d and all real nodes 32a to 32d. These collected node internal information and statistical information are output to the confirmation unit 27A.

The evaluation result confirmation unit 27A evaluates the input NW model performance measurement information, node internal information, and statistical information of the NW model 21A for evaluation point k (for example, k=2nd) on the entire evaluation space of the NW model 21A. result. At the same time, the confirmation unit 27A checks the input real NW performance measurement information of all the real nodes 32a to 32d and the node internal information and statistical information of all the node models 21a to 21d and all the real nodes 32a to 32d of the real NW 31A. Assume that the evaluation result is related to the evaluation point k (for example, k=2nd) on the entire evaluation space.

The confirmation unit 27A associates the evaluation result of the NW model 21A and the evaluation result of the actual NW 31A with respect to the evaluation point k (for example, k=2nd) on the entire evaluation space and outputs them to the recording unit 28A. The recording unit 28A records each evaluation result of the NW model 21A and the actual NW 31A related to the evaluation point k (for example, k=2nd) in the entire evaluation space in association with each other.

Based on the evaluation results of the NW model 21A and the actual NW 31A that are associated and recorded in the recording unit 28A, the evaluation result analysis unit 29A performs the above-described analysis for estimating the latest parameters as follows.

That is, the analysis unit 29A searches for an evaluation point k in the entire evaluation space at which the absolute value of the difference between the evaluation result of the NW model 21A and the evaluation result of the actual NW 31A is equal to or less than a predetermined threshold value, and Let the parameter corresponding to k be the nearest parameter. This recent parameter is output to and stored in the policy section 30A.

However, in the latest parameter estimation by the analysis unit 29A, it is determined whether or not the absolute value of the difference is equal to or less than the threshold value for each evaluation point k in the entire evaluation space. Without waiting for the completion of the evaluation for the evaluation point k=L, the parameter at the point of time below the threshold may be used as the latest parameter. If the absolute value of the above difference does not fall below the threshold even after completing the evaluation for all evaluation points k=L on the full-value space, additional evaluation points on the full-value space are set to expand the full-value space. may

In addition, when the NW model performance measurement information and the actual NW performance measurement information are input from the performance measurement unit 24A, the confirmation unit 27A checks the evaluation point k (for example, j=2nd) on the entire evaluation space of the NW model 21A. The evaluation control unit 25A is notified of the end of the evaluation. At the time of this notification, the evaluation for the second evaluation point on the entire evaluation space ends.

When the evaluation control unit 25A receives the notification that the evaluation for the evaluation point k (for example, j=2nd) on the entire evaluation space has been completed, the evaluation control unit 25A selects the next evaluation point k=3rd on the entire evaluation space from the policy unit 30A. Read parameters to evaluate . In addition, the evaluation control unit 25A terminates the evaluation process when receiving the notification that the evaluation for all the evaluation points k=Lth in the entire evaluation space has been completed.

<Operation of Second Embodiment>
Next, the second communication network model parameter setting support processing by the support device 10A of the second embodiment will be described with reference to the flowchart shown in FIG.

However, in the policy section 30A, the latest parameters of the node models 21a to 21d are stored as initial parameters in the same manner as the policy section 30 (FIG. 1) described above, and the priority when searching for parameters is set to each Assume that it is stored in association with a parameter.

In step S11 shown in FIG. 4, the evaluation control unit 25A initializes k=1. k corresponds to an evaluation point on the entire evaluation space and is varied from k=1 to L one by one.

In step S12, the evaluation control unit 25A reads from the policy unit 30A the initial parameters corresponding to the evaluation point k=1 on the evaluation space for all of the node models 21a to 21d, and also reads the traffic pattern from the DB unit 26A. The read parameters are output to the setting section 23A, and are set in the respective node models 21a to 21d of the NW model 21A by the setting section 23A. Also, the traffic pattern read by the evaluation control section 25A is output to the load testing section 22A.

In step S13, the load test unit 22A inputs the input traffic pattern to the NW model 21A and the actual NW 31A for a predetermined period of time.

In step S14, the performance measurement unit 24A measures performance such as throughput, delay, and packet discard rate during operation of the NW model 21A and the real NW 31A to which the traffic pattern is input. The measured NW model performance measurement information and actual NW performance measurement information are output to the confirmation unit 27A.

In step S15, the collection unit 20A collects node internal information and statistical information of all node models 21a to 21d and all real nodes 32a to 32d. The collected node internal information and statistical information are output to the confirmation unit 27A.

In step S16, the confirmation unit 27A compares the input NW model performance measurement information, node internal information, and statistical information of the NW model 21A with the evaluation result of the evaluation point k=1 on the entire evaluation space of the NW model 21A. do. At the same time, the confirmation unit 27A checks the input real NW performance measurement information of the real NW and the node internal information and statistical information of all the node models 21a to 21d and all the real nodes 32a to 32d on the whole evaluation space of the real NW 31A. is the evaluation result for the evaluation point k=1. Confirmation unit 27A associates the evaluation results of both NW model 21A and actual NW 31A and outputs them to recording unit 28A. Further, when the NW model performance measurement information and the actual NW performance measurement information are input, the confirmation unit 27A notifies the evaluation control unit 25A of the end of the evaluation for the evaluation point j=1st in the entire evaluation space. At the time of this notification, the evaluation for the first evaluation point on the entire evaluation space ends.

In step S17, the evaluation control section 25A determines whether or not k=L upon receiving the notification from the confirmation section 27A. In this case, since k is not L (No), returning to step S12, the evaluation control unit 25A selects the evaluation point k=k+1 (here, k=2) on the entire evaluation space from the policy unit 30A. Read parameters. After that, the processing of steps S13 to S17 is repeated.

After that, in step S17, the evaluation control unit 25A determines whether or not k=L, and if the result is k=L (Yes), the process proceeds to step S18.

In step S18, when the evaluation control unit 25 notifies the analysis unit 29A of the completion of the evaluation of all evaluation points k=L in the entire evaluation space related to the NW model 21A, the analysis unit 29A An analysis for estimating the latest parameters, which will be described later, is performed from each evaluation result of k=1 to L as follows.

That is, the analysis unit 29A searches for an evaluation point k in the entire evaluation space at which the absolute value of the difference between the evaluation result of the NW model 21A and the evaluation result of the actual NW 31A is equal to or less than a threshold value, and corresponds to the evaluation point k. Let the parameters be recent parameters. This recent parameter is output to and stored in the policy section 30A. This completes the second communication NW model parameter setting support process.

<Effects of Second Embodiment>
Effects of the support device 10A according to the second embodiment of the present invention will be described.

The support device 10A includes a NW model 21A, a setting unit 23A, a load test unit 22A, a performance measurement unit 24A, a collection unit 20A, a confirmation unit 27A, and an analysis unit 29A.

The NW model 21A corresponds to a real NW 31A in which real nodes 32a to 32d of different node types as communication devices are connected by NW (Network) and deployed. 21d are connected.

The setting unit 23A sets parameters for each of the node models 21a to 21d corresponding to each parameter of each of the real nodes 32a to 32d of the real NW 31A. The load testing unit 22A inputs traffic patterns that change node loads to the NW model 21A and the real NW 31A.

The performance measurement unit 24A measures the performance of each of the NW model 21A and the actual NW 31A that operate with traffic pattern inputs, and obtains performance measurement information for each. The collection unit 20A collects node internal information and statistical information of all node models 21a to 21d and all real nodes 32a to 32d.

The verification unit 27A checks the input performance measurement information of each of the NW model 21A and real NW 31A and the node internal information and statistical information of each of all node models 21a to 21d and all real nodes 32a to 32d to the NW model 21A and real NW 31A. This is the evaluation result of the evaluation points on the entire evaluation space in NW31A.

The analysis unit 29A sets the parameters for each of the node models 21a to 21d corresponding to the evaluation points at which the absolute value of the difference between the evaluation results of the NW model 21A and the real NW 31A is equal to or less than a predetermined threshold, to all of the real nodes. The most recent parameters closest to the parameters 32a to 32d are estimated.

According to this configuration, it is possible to estimate and set parameters (recent parameters) that are closest to all the real nodes 32a to 32d that make up the real NW 31A for all the node models 21a to 21d that make up the NW model 21A.

In addition, each recent parameter related to the evaluation point j on the evaluation space of each node type estimated by the analysis unit 29 described in the first embodiment is initialized to each of the node models 21a to 21d described in the second embodiment. Set as a value.

Furthermore, a higher priority is given to the parameter related to the evaluation point k on the entire evaluation space in the NW model 21A and the actual NW 31A described in the second embodiment, as the sensitivity of the parameter having a greater influence on node processing increases. The analysis unit 29 evaluates whether or not the absolute value of the difference between the evaluation results of the NW model 21A and the actual NW 31A is equal to or less than the threshold value in order of the evaluation point k related to the parameter with the highest priority, and estimates the latest parameter. may be performed.

<Program>
Also, a program executed by a computer according to the present embodiment (first embodiment) will be described. The computer corresponds to a real NW 31 in which a plurality of real nodes 32a as communication devices are network-connected and deployed, and has a network model 21 to which a node model 21a corresponding to the real nodes 32a is connected. Consider the device 10 (see FIG. 1).

This program provides the computer with means for setting node model parameters corresponding to the parameters of the real node 32a in the node model 21a, and means for inputting traffic patterns that change the node load to the node model 21a and the real node 32a. , the performance measuring unit 24 measures the performance of each of the node model 21a alone and the real node 32a alone, which are operated by the input of the traffic pattern, to obtain performance measurement information of each; According to the means for collecting each node internal information and statistical information, the performance measurement information of each node model 21a single unit and real node 32a single unit, and the node internal information and statistical information, the evaluation points in the evaluation space in one node type Means for acquiring such node evaluation results and recording them in the recording unit 28, corresponding to the evaluation point that minimizes the absolute value of the difference between the evaluation results of the single node model 21a and the single real node 32a recorded in the recording unit 28 The parameters for the node model function as means for estimating the latest parameters that are closest to the parameters of the real node 32a.

According to this program, similar to the effects of the support device 10 described above, it is possible to estimate and set the parameters closest to the real nodes 32a that make up the real NW 31 (most recent parameters) in the node models 21a that make up the NW model 21. . However, the program is stored in a storage medium, and a CPU (Central Processing Unit) reads the program from the storage medium and executes it.

<effect>
(1) A NW model to which a node model corresponding to the real node is connected, corresponding to a real NW in which real nodes of different node types as communication devices are connected to each other and deployed; A setting unit for setting corresponding node model parameters to the node model; a load test unit for inputting a traffic pattern that changes the node load to the node model and the real node; and an input of the traffic pattern. a performance measuring unit that measures the performance of each of the node model unit and the real node unit to obtain performance measurement information of each, and collects node internal information and statistical information of each of the node model unit and the real node unit. Confirmation that the collecting unit and the recording unit record the performance measurement information, node internal information, and statistical information of each of the node model unit and the real node unit as node evaluation results related to evaluation points in the evaluation space for one node type. and the parameter for the node model corresponding to the evaluation point with the minimum absolute value of the difference between the evaluation results of the single node model and the single real node recorded in the recording unit, which is closest to the real node. A communication network model parameter setting support device comprising parameters and an analysis unit for estimating the parameters.

According to this configuration, it is possible to estimate and set the parameters (most recent parameters) that are closest to the real nodes that make up the real NW in the node models that make up the NW model.

(2) a NW model in which each node model corresponding to each real node is connected; A setting unit for setting, for each node model, parameters for each node model corresponding to each parameter for each real node of the NW, and a load test for inputting a traffic pattern that changes the node load to the NW model and the real NW. a performance measurement unit that measures the performance of each of the NW model and the real NW that operate on the input of the traffic pattern to obtain performance measurement information of each; A collection unit for collecting node internal information and statistical information of each of all real nodes, performance measurement information of each of the NW model and the real NW, and node internal information and statistics of each of the all node models and all real nodes A confirmation unit that records information in a recording unit as evaluation results related to evaluation points on all evaluation spaces in the NW model and the actual NW; an analysis unit for estimating the parameter for each of the node models corresponding to the evaluation point where the absolute value of the difference is equal to or less than a predetermined threshold as the nearest parameter closest to all the real nodes. It is a communication network model parameter setting support device.

According to this configuration, it is possible to estimate and set parameters (closest parameters) that are closest to all the real nodes that make up the real NW for all the node models that make up the NW model.

(3) setting each recent parameter related to the evaluation point in the evaluation space of each node type estimated by the analysis unit of the communication network model parameter setting support device according to claim 1 as an initial value in each of the node models; Further, a higher priority is given to the parameters related to the evaluation points on the entire evaluation space in the NW model and the real NW, the higher the sensitivity of the parameter having a large influence on the node processing, and the analysis unit prioritizes The most recent parameter is estimated by evaluating whether or not the absolute value of the difference between the evaluation results of the NW model and the actual NW is equal to or less than a threshold in the order of the evaluation points related to the parameter with the highest degree. The communication network model parameter setting support device according to (2) above.

In addition, the specific configuration can be changed as appropriate without departing from the gist of the present invention.

10, 10A Communication network model parameter setting

support device

20, 20A Node information collection unit (collection unit)
21, 21A NW model 21a to

21d node model

22, 22A

load test section

23, 23A parameter setting section (setting section)
24, 24A

performance measurement section

25, 25A

evaluation control section

26, 26A evaluation space

information DB section

27, 27A evaluation result confirmation section (confirmation section)
28, 28A Evaluation result recording unit (recording unit)
29, 29A Evaluation result analysis unit (analysis unit)
30, 30A Parameter policy section (policy section)
31, 31A Actual NW
32a-32d real nodes

Claims

a NW model to which a node model corresponding to the real node is connected, corresponding to a real NW in which real nodes of different node types as communication devices are connected to each other and deployed;
a setting unit that sets parameters for a node model corresponding to the real node to the node model;
a load testing unit that inputs a traffic pattern that changes the node load to the node model and the real node;
a performance measuring unit that measures the performance of each of the node model unit and the real node unit that operate with the input of the traffic pattern and obtains performance measurement information of each;
a collection unit that collects node internal information and statistical information of each of the node model unit and the real node unit;
a confirmation unit that records, in a recording unit, performance measurement information, node internal information, and statistical information of each of the node model unit and the real node unit as node evaluation results related to evaluation points in an evaluation space for one node type;
The parameter for the node model corresponding to the evaluation point with the minimum absolute value of the difference between the evaluation results of the single node model and the single real node recorded in the recording unit is estimated as the nearest parameter closest to the real node. and a communication network model parameter setting support device.
a NW model in which real nodes of different node types as communication devices correspond to a real NW (Network) connected and deployed, and each corresponding node model is connected to each real node;
a setting unit for setting, for each node model, a parameter for each node model corresponding to each parameter for each real node of the real NW;
a load testing unit that inputs traffic patterns that change node loads to the NW model and the real NW;
a performance measuring unit that measures the performance of each of the NW model and the real NW that operate with the input of the traffic pattern and obtains performance measurement information of each;
a collection unit that collects node internal information and statistical information of all node models of the NW model and all real nodes of the real NW;
Performance measurement information of each of the NW model and the real NW, node internal information and statistical information of all node models of the NW model and all real nodes of the real NW, and all evaluations in the NW model and the real NW a confirmation unit that records in a recording unit as an evaluation result related to spatial evaluation points;
The parameter for each node model corresponding to the evaluation point at which the absolute value of the difference between the evaluation results of the NW model and the actual NW recorded in the recording unit is equal to or less than a predetermined threshold is set to all the real nodes. A communication network model parameter setting support device, comprising: an analysis unit for estimating the nearest recent parameter.
Each recent parameter related to the evaluation point in the evaluation space of each node type estimated by the analysis unit of the communication network model parameter setting support device according to claim 1 is set as an initial value in each node model,
A higher priority is given to the parameters related to the evaluation points on the entire evaluation space in the NW model and the real NW, the higher the parameter sensitivity that has a greater influence on the node processing, and the analysis unit assigns a higher priority. estimating the latest parameter by evaluating whether or not the absolute value of the difference between the evaluation results of the NW model and the actual NW is equal to or less than a threshold in the order of the evaluation points related to the parameters; Item 3. A communication network model parameter setting support device according to item 2.
A communication network model parameter setting support method by a communication network model parameter setting support device,
The communication network model parameter setting support device,
A NW model corresponding to a real network in which real nodes of different node types as communication devices are connected to each other and deployed, and to which a node model corresponding to the real node is connected,
setting a parameter for a node model corresponding to the real node to the node model;
inputting a traffic pattern that changes node load to the node model and the real node;
measuring the performance of each of the node model unit and the real node unit operating with the input of the traffic pattern to obtain performance measurement information of each;
collecting node internal information and statistical information for each of the node model unit and the real node unit;
a step of recording performance measurement information, node internal information, and statistical information of each of the node model unit and the actual node unit as node evaluation results related to evaluation points in the evaluation space of one node type in a recording unit;
The parameter for the node model corresponding to the evaluation point with the minimum absolute value of the difference between the evaluation results of the single node model and the single real node recorded in the recording unit is estimated as the nearest parameter closest to the real node. A communication network model parameter setting support method, characterized by:
A communication network model parameter setting support method by a communication network model parameter setting support device,
The communication network model parameter setting support device,
A NW model corresponding to a real NW in which real nodes of different node types as communication devices are connected to each other and deployed, and each node model corresponding to each real node is connected,
setting parameters for each node model corresponding to each real node of the real NW;
inputting a traffic pattern that changes the node load to the NW model and the real NW;
measuring the performance of each of said NW model and said real MW operating on the input of said traffic pattern to obtain respective performance measurement information;
collecting node internal information and statistical information of each of all node models of said NW model and all real nodes of said real NW;
Performance measurement information of each of the NW model and the real NW, node internal information and statistical information of all node models of the NW model and all real nodes of the real NW, and all evaluations in the NW model and the real NW a step of recording in a recording unit as a node evaluation result related to spatial evaluation points;
The parameter for each node model corresponding to the evaluation point at which the absolute value of the difference between the evaluation results of the NW model and the actual NW recorded in the recording unit is equal to or less than a predetermined threshold is set to all the real nodes. A communication network model parameter setting support method, characterized by performing a step of estimating the nearest recent parameter.
Each recent parameter related to the evaluation point in the evaluation space of each node type estimated by the communication network model parameter setting support method according to claim 4 is set as an initial value in each node model,
The communication network model parameter setting support device,
A higher priority is given to the parameters related to the evaluation points on the entire evaluation space in the NW model and the real NW, the higher the sensitivity of the parameter having a large influence on the node processing, and the evaluation points related to the parameters with higher priority. The step of estimating the latest parameters by evaluating whether or not the absolute value of the difference between the evaluation results of the NW model and the actual NW is equal to or less than a threshold in the order of 6. The communication network model parameter setting support method according to 5.
A program for causing a computer to function as the communication network model parameter setting support device according to any one of claims 1 to 3.