JP3430930B2 - Method and apparatus for estimating traffic in packet switched network - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a traffic estimation method and device in a packet switching network for simulating traffic in a packet switching network composed of a plurality of route control devices.

[0002]

2. Description of the Related Art In a small-scale packet switching network such as an intranet (hereinafter, also simply referred to as "network"), the number of elements (for example, host computer, router and line) constituting the network is small. Therefore, in such a small network, these constituent elements can be defined in advance. Furthermore, in such a small-scale network, it is possible to estimate the change in usage to some extent.

Conventionally, many methods for simulating the traffic of such a small network have been proposed. In the simulation,
Usually, a network model is created using software that applies queues and probabilities. Specifically, for example, a packet signal is transmitted on each host according to a specific probability distribution, and the communication buffer on each router is considered as a queue. Then, based on the bandwidth of the line and the processing capability derived from the specifications of the router, the average processing time for each packet signal is obtained in multiple stages for each hop using queuing theory.

[0004]

By the way, in a large-scale open packet switching network such as the Internet, the organization of network elements such as routers and lines is divided into many organizations. Therefore, in a large-scale network, it is difficult to grasp the bandwidth of each router or line and the link status at one place.

In particular, the networks managed by the organizations are not open to the public between independent organizations (for example, between commercial providers). For this reason, it is difficult to grasp information about the usage status of the entire network, such as where, how services are used, and how much traffic has occurred in the network.

Further, in a large scale network,
It is difficult to understand the information on the usage status and the usage status changes drastically. Therefore, it is practically impossible to create a model of a large-scale network using a conventional probabilistic model using a queue or the like and perform a simulation on traffic.

Therefore, conventionally, for a large-scale network, the network model has not been used to predict traffic and to be useful for route control. In addition, large-scale networks are generally designed and controlled based on experience-based intuition.

The present invention has been made to solve the above problems, and an object of the present invention is to provide a traffic estimation technique in a packet switching network which simulates a track of the packet switching network.

[0009]

(Means for Solving the Problems) (Traffic Estimating Method in Packet Switching Network) In order to achieve this object, a traffic estimating method in a packet switching network according to the present invention comprises
In a packet-switched network consisting of a plurality of links, a data set of round-trip paths and round-trip times of packet signals between a fixed observation point and a plurality of points is obtained, and a node equal to all the links A neural network having an input layer having one and an output layer, a teacher signal created based on the data set, and a vector having a dimension equal to the number of all the links, A vector of values obtained by normalizing the value of the component corresponding to the link with the maximum coefficient value is input as input data, and the neural network performs learning based on this teacher signal and input data to create a network model. In the network model, a vector of dimension equal to the number of all the links,
Input test data in which the value of the component corresponding to the link on the arbitrary path is a value obtained by dividing the number of passages of the link by the maximum number of passages, and based on the output of the network model, the packet switching network This is a method for estimating traffic on an arbitrary route.

As described above, according to the traffic estimation method in the packet switching network of the present invention, the data set of the round trip route and the round trip time is collected at the fixed observation point. Then, learning by knowledge processing is performed based on this data set. Then, as a result of learning by this knowledge processing, information about the correlation between the traffic of each link and the traffic of the entire network can be obtained.

Specifically, in learning by knowledge processing, a neural network having an input layer having nodes equal in number to all links and one output layer is used, and a vector having a dimension equal to the number of all links is used. In addition, the vector of the coefficient value divided by the maximum coefficient value (the maximum value of the coefficient values) of the component corresponding to each link on the round-trip path is used as the input data, and the teacher time is the round-trip time. And input to the neural network.

When the input data based on the data set which is the actual measurement value is input, the neural network compares the output with the value of the teacher signal as the sample value which is the actual measurement value to correct the neural network. Learn by and create a network model.

Further, in estimating the traffic of an arbitrary route, the network model uses a vector having a dimension equal to the number of all links, and passing a value of a component corresponding to a link on the arbitrary route through the link. Input the test data that is the number of times divided by the maximum number of times of passage (the maximum number of times of passage), and as the output of the network model, the expected communication time corresponding to the estimated required time of any route. Ask.

A packet signal passing through a link on the network is usually transferred from another link adjacent to the link. The adjacent link is
It is also adjacent to other links. Therefore, the traffics of the individual links configuring the network have some correlation with each other. That is, traffic on one link affects traffic on another link. And this correlation is acquired as a result of learning by knowledge processing. Then, as a result of the learning, for example, when the traffic of a certain link rises, it is possible to estimate the influence occurring on the traffic of another link.

The data set which is the target of the knowledge processing is only a part of the information indicating the usage status of the network. That is, according to the present invention, it is possible to estimate the traffic of an arbitrary route by using the result of the knowledge processing without grasping all the constituent elements and all the usage states of the network.

Further, in the traffic estimation method in the packet-switching network of the present invention, preferably, the correlation of traffic is a correlation between the time required for the packet signal to pass through the link and the round-trip time. Is desirable.

The time required for the link is affected by the traffic volume (communication volume) in the network. Therefore,
The correlation between the required time and the round trip time of a certain link reflects the correlation between the traffic volume of the link and the traffic volume of the round trip route.

In implementing the present invention, preferably, a variable corresponding to the required time of the link on the round-trip route is provided, the number of times the link passes is used as a coefficient, and the sum of the variables multiplied by the coefficient is used as the round-trip time. It is advisable to establish the above simultaneous equations on the basis of the data set and to obtain a set of solutions of the simultaneous equations by learning.

By the way, the number of data sets is usually smaller than the number of links in the network. Therefore, the number of simultaneous equations corresponding to each data set is smaller than the number of variables corresponding to the time required for each link. Therefore,
Normally, the solution of simultaneous equations cannot be uniquely obtained.

All solutions (time required for each link)
Has a constraint that it takes a value of 0 or more. Therefore, it is possible to solve this simultaneous equation as a linear programming problem using this constraint. However, as a linear programming problem, for example, when a solution is obtained by the simplex method, only the value of the solution that can be the maximum value (link that can maximize the required time) can be calculated from all the solution sets.

On the other hand, the data set on which the simultaneous equations are based is merely a measurement of the state of the network at a certain time. On the other hand, network usage changes drastically. Moreover, the measured values of the dataset usually include errors. Therefore, it is not practical to calculate only the maximum possible solution for a linear programming problem.

That is, when only one solution is calculated, it is not possible to obtain another solution which has a high probability of being present other than the calculated solution and which takes a long time. Moreover, the calculated solution itself may be incorrect due to an error.
Therefore, rather than calculating one solution, it is practically desirable to estimate a candidate group of solutions (rings that may take a long time) that can take a large value with a certain probability by knowledge processing. Therefore, in this invention,
Even for problems that can be solved as linear programming problems,
Apply knowledge processing.

(Traffic Estimating Device in Packet Switching Network Network) Further, the traffic estimating device in the packet switching network network of the present invention is provided at a fixed observation point connected to the packet switching network network configured by connecting a plurality of links. And a round-trip route detection unit that detects a round-trip route of a packet signal between a fixed observation point and a plurality of points,
A round-trip time detecting unit that detects a round-trip time required for the packet signal to be transferred through the round-trip route, a storage unit that stores a data set of the round-trip route and the round-trip time, and all links based on the data set. Is a vector with a dimension equal to the number of, and the value of the component corresponding to each link on the round-trip path is the value obtained by dividing the number of passes of the link by the maximum number of passes, and the value of the component corresponding to other links is zero The input data and the teacher signal creation unit that creates the teacher signal corresponding to the round-trip time, and the vector of the same dimension as the input data, corresponding to each link on any path The value of the component to be used is the value obtained by dividing the number of passes of the link by the maximum number of passes, and a vector with the value of the component corresponding to another link being zero is created as test data. And Todeta creating unit, as well as learning a correlation between the duration and round-trip time for each link based on the input data and the teacher signal, based on the learning result,
And a neural network that outputs an expected transit time of an arbitrary route designated by the test data.

As described above, according to the traffic estimation device in the packet-switching network of the present invention, the round-trip route detecting unit detects the round-trip route and the round-trip time detecting unit detects the round-trip time at the fixed observation point. Then, the data set of the detected round trip route and round trip time is stored in the storage unit.

The input data / teacher signal generation section
Input data and teacher signals are created based on the data set and input to the neural network. Then, as a result of the learning, the neural network obtains information about the correlation between the traffic of each link and the traffic of the entire network.

Further, the test data creating section creates test data for designating an arbitrary route and inputs it to the neural network. As a result, the neural network outputs the expected transit time of the designated route of the test data. Therefore, the traffic of any route can be obtained.

[0027]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. The present invention is not limited to the illustrated example. In this embodiment, an example will be described in which the first host computer (first host) 10 in the network 100 shown in FIG. 1 is used as a fixed observation point.

The processing of the traffic estimation method in the packet switching network according to this embodiment is executed under the control of the program in the first host 10 computer. This program, for example,
It is provided by a recording medium. As the recording medium, for example, a magnetic disk, a semiconductor memory, or any other computer-readable medium can be used.

The network 100 includes a first router 20,
Second router 22, third router 24 and fourth router 26
It is composed by. Then, in the first router 20,
A first host computer (first host) 10 is connected. A second host computer (second host) 12 is connected to the second router 22. A third host computer (third host) is connected to the third router 24. A fourth host computer (fourth host) 16 is connected to the fourth router 26.

The first router 20 and the second router 22 are connected by a link A. The second router 22 and the third router 24 are connected by a link B. The third router 24 and the fourth router 26 are connected by a link C. The fourth router 26 and the first router 20 are connected by a link D.
The fourth router 26 and the second router 22 are connected to each other by link E
Connected by.

The first host 10 is provided with a traffic estimation device in the packet switching network as a fixed observation point. Here, FIG. 2 shows a block diagram of the traffic estimation device 110 in the packet switching network. As shown in FIG. 2, the traffic estimation device 110 in this packet-switching network includes a round-trip route detection unit 111, a round-trip time measurement unit 112 as a round-trip time detection unit, a database 113 as a storage unit, and input data. Teacher signal creation unit 114
And a test vector creation unit 115 as a test data creation unit, and a neural network (neural net) 116.

Then, the round-trip route detecting section 111 is arranged so that the fixed observation point and a plurality of points, for example, the first host 10 and the second point.
-Detects the round-trip path of each packet signal to and from the fourth hosts 12, 14 and 16, respectively.

Here, the first host 10 and the third host 14
An example of detecting a round-trip path between and will be described. In this embodiment, first, a plurality of search packet signals (traffic data collection packets) are created. The search packet signal sets the first host 10 as a sender as a destination and the third host 14 as a waypoint.

Next, a plurality of search packet signals are set by sequentially increasing the survival time of the search packet signals and transmitted from the first host 10 in sequence. Then, when the time to live of the search packet signal expires, an error packet signal transmitted from the router or the third host 14 on the way is received. Then, a route that passes through the router and the third host 14 in the order in which the error packet signal is transmitted is detected as a round trip route. Further, other round trip routes are detected in the same manner.

Further, the round-trip time measuring unit 112 of the traffic estimation device 110 in the packet switching network.
Detects the round-trip time required for a packet signal to be transferred along the round-trip path.

The database 113 also stores a data set of detected round trip routes and round trip times, which is traffic data.

Further, the input data / teacher signal creating section 114
Creates input data and a teacher signal based on the data set. This input data is a vector having a dimension equal to the number of all links, and is a vector in which the value of the component corresponding to each link on the round-trip route is the value obtained by dividing the number of times of passing the link by the maximum number of passing. . Therefore, the value of the component corresponding to the link other than the round trip route is zero. The teacher signal has a value corresponding to the round trip time of the round trip route corresponding to the input data.

Further, the test vector creating section 115 creates a test vector as test data. This test vector is a vector of the same dimension as the input data, and the value of the component corresponding to each link on any route is the value obtained by dividing the number of times of passing the link by the maximum number of times of passing,
It is a vector in which the value of the component corresponding to another link is zero. That is, the test data can specify an arbitrary route of the network.

The test vector creation unit 115 is
Information representing the correspondence between each link and each component of the vector is transferred from the database 113.

Further, the neural network 116 is
The correlation between the required time and the round trip time of each link is learned based on the input data and the teacher signal. Then, this neural network 116, based on the learning result,
Outputs the expected transit time of an arbitrary route specified by test data.

Next, with reference to FIG. 3, a specific example of the traffic estimation method in the packet switching network, that is, the operation of the traffic estimation device in the packet switching network will be described. Figure 3
3 is a flow chart for explaining a traffic estimation method in a packet switching network.

In creating the network model, first, the round trip path of the packet signal between the fixed observation point and a plurality of points is detected (S1 in FIG. 3). Here, first
A first round trip path between the first host 10 and the second host 12 is detected. The first round-trip route starts from the first host 10 and goes to the first router 20, the second router 22 and the second host 1.
It is assumed that the route is such that the route 2 sequentially passes through the second router 22, the fourth router 26, and returns to the first host 10. This first round-trip route is represented by a link as “link A → link B → link D”.

Next, the time required for the packet signal to be transferred through the first round trip path is obtained as the first round trip time (S2 in FIG. 3). Here, the first round trip time is "10 seconds"
It was.

Next, it is judged whether or not the necessary data sets of the round trip route and the round trip time have been obtained (S in FIG. 3).
3). The reference amount of judgment can be set in advance.
In this embodiment, the data set of the round trip path and the round trip time between the first host 10 and each of the other hosts is obtained, and it is determined that the data set has been obtained as much as necessary.

Then, the first host 10 and the third host 14
The second round trip route to and from is "Link D → Link E → Link B → Link C → Link E
→ Link A ”is detected. In addition, the second time, which is the time required for the second round trip route,
The round trip time is detected as "50 seconds".

Further, the third round-trip path between the first host 10 and the fourth host 16 is detected as "link D → link C → link B → link A". In addition, the third time, which is the time required for the third round trip route,
The round trip time is detected as "40 seconds". The data set of the round trip route and the round trip time thus detected is stored in the database 113.

Next, a variable corresponding to the required time of the link on the round trip route is provided. That is, the variable a is given as the time required for the link A. Similarly, the variable b is given to the link B, the variable c is given to the link C, the variable d is given to the link D, and the variable e is given to the link E.

Since the sum of the times required for each link on the round-trip route multiplied by the number of times the link has passed is equal to the round-trip time, simultaneous equations shown in the following equations (1) to (3) hold. .

That is, for the first round trip path, a + d + e = 10 (1) holds. Further, for the second round-trip path, a + b + c + d + 2e = 50 (2) holds. However, the coefficient "2" of the variable e is the link E
Represents the number of passes through. That is, in the second round trip path, the link E is passed twice. For the third round-trip route, a + b + c + d = 40 (3) holds.

The set of solutions of the simultaneous equations consisting of the above equations (1) to (3) holds the correlation between the time required for the packet signal to pass through each link and the round-trip time. . In other words, this correlation is a traffic correlation between the traffic of the link and the traffic of the entire packet switching network. Then, this correlation is acquired as a learning result of knowledge processing.

Therefore, in order to learn the simultaneous equations by the neural network, input data is created and input to the neural network together with the teacher signal (S4 in FIG. 3).

The input data is a vector having a dimension equal to the number of all links, and the value of the component corresponding to each link on the reciprocating route is obtained by dividing each coefficient by the maximum coefficient value (maximum coefficient value). Is a vector that is normalized with. Here, the variables a to e are respectively associated with the first to fifth components of the vector. Then, the round trip time is used as a teacher signal.
The teacher signal is also normalized by dividing it by the maximum coefficient value. Specifically, for the first round-trip path, input data: (1,0,0,1,1) teacher signal: 10 is input.

For the second round trip path, the input data: (0.5, 0.5, 0.5, 0) is obtained by dividing each coefficient value and the teacher signal by the maximum coefficient (maximum number of passages) 2. .5, 1) Input the teacher signal: 25 (= 50/2).

Regarding the third round trip route, Input data: (1,1,1,1,0) Teacher signal: 40 Enter.

Further, here, as the neural network, a neural network provided with an input layer having nodes equal in number to all links and one output layer is used. In the neural network, the maximum value of the teacher signal is set to 1, and each teacher signal is standardized and used.

Then, the neural network learns the neural network by comparing the output of the input data with the teacher signal and correcting the variables (S5 in FIG. 3). As a result of this learning, a neural net corresponding to the network model is obtained. Then, the result of this learning holds the magnitude of the influence of the output of each component of the input data, that is, the information on the correlation between each link and the round trip time. Therefore, the neural network obtained as a result of this learning corresponds to the network model.

Next, using the result of learning, the traffic of an arbitrary route of the packet switching network is estimated (S6 in FIG. 3). Therefore, here, the test vector creation unit 115 creates a test vector.
This test vector is a unit vector. And
The test vector is a vector of the same dimension as the number of all links, and the value of the component corresponding to the link on any route is the value obtained by dividing the number of times of passing the link by the maximum number of times of passing, and It is a vector in which the value of the corresponding component is zero.

Then, this test vector is input to the learned neural network. Then, as the output of the neural network, the expected communication time corresponding to the estimated time required for the packet signal to pass through the arbitrary route designated by the test vector is obtained.
As this expected communication time, the estimated required time itself may be output, or data for which the estimated required time can be easily obtained, for example, an integer multiple of the estimated required time may be output. good.

In the above-described embodiment, the example in which the present invention is configured under the specific condition has been described, but the present invention can be variously modified. For example, in the above-described embodiment, an example in which a neural network is used as knowledge processing has been described, but in the present invention,
Knowledge processing is not limited to this. As the knowledge processing, for example, a genetic algorithm may be used.

[0060]

As described above in detail, according to the present invention, a data set of a round trip route and a round trip time is collected at a fixed observation point. Then, learning by knowledge processing is performed based on this data set. Then, as a result of learning by this knowledge processing, information about the correlation between the traffic of each link and the traffic of the entire network is obtained. Furthermore, the traffic of an arbitrary route is estimated using the result of knowledge processing.

[Brief description of drawings]

FIG. 1 is a schematic diagram of the Internet for explaining this embodiment.

FIG. 2 is a block diagram for explaining a traffic estimation device in the packet-switched network of this embodiment.

FIG. 3 is a flow chart for explaining a traffic estimation method in a packet switching network according to this embodiment.

[Explanation of symbols]

10 first host computer (first host) 12 second host computer (second host) 14 third host computer (third host) 16 fourth host computer (fourth host) 20 first router 22 second router 24 3 router 26 4th router 100 network 110 traffic estimation device in packet switching network 111 round-trip route detection unit 112 round-trip time measurement unit 113 database 114 input data / teacher signal creation unit 115 test vector creation unit 116 neural network

─────────────────────────────────────────────────── ─── Continued Front Page (58) Fields surveyed (Int.Cl. ⁷ , DB name) H04L 12/56 G06F 15/18

Claims (4)

(57) [Claims] 1. In a packet-switched network consisting of a plurality of links, a data set of a round trip path and a round trip time of a packet signal between a fixed observation point and a plurality of points is obtained, A neural network having an input layer having nodes equal to the number of links and one output layer, a teacher signal created based on the data set, and a vector having a dimension equal to the number of all the links, A vector of values obtained by normalizing the value of the component corresponding to each link on the round-trip path with the maximum coefficient value is input as input data, and the neural network performs learning based on the teacher signal and the input data, whereby the network model is obtained. To this network model, a vector of dimension equal to the number of all said links, Input test data in which the value of the component corresponding to a link on an arbitrary route is a value obtained by dividing the number of passages of the link by the maximum number of passages, and based on the output of the network model, the packet switching network A method of estimating traffic in a packet-switched network, characterized by estimating traffic of an arbitrary route. 2. The traffic estimation method in a packet-switched network according to claim 1, wherein the correlation of the traffic is a correlation between a time required for the packet signal to pass through the link and the round-trip time. A method for estimating traffic in a packet-switched network characterized by the relation. 3. The traffic estimation method in a packet-switched network according to claim 2, wherein a variable corresponding to the time required for the link on the round-trip route is provided, and the number of times the link has passed is used as a coefficient, A method of estimating traffic in a packet-switching network, characterized in that a simultaneous equation with the sum of the variables multiplied as the round-trip time is established based on the data set, and a set of solutions of the simultaneous equation is obtained by the learning. . 4. A round trip path of a packet signal between a fixed observation point and a plurality of points, which is provided at a fixed observation point connected to a packet switching network constituted by connecting a plurality of links. A round-trip path detecting section; a round-trip time detecting section that detects a round-trip time required for the packet signal to be transferred through the round-trip path; a storage section that stores a data set of the round-trip path and the round-trip time; Based on the data set, a vector having a dimension equal to the number of all the links, and the value of the component corresponding to each link on the round-trip path is a value obtained by dividing the number of times of passing of the link by the maximum number of times of passing, Input data / teacher signal creation for creating a vector in which the value of the component corresponding to another link is zero as input data and creating a teacher signal corresponding to the round trip time And a vector of the same dimension as the input data, in which the value of the component corresponding to each link on an arbitrary path is the value obtained by dividing the number of times of passing the link by the maximum number of times of passing as test data. A test data creating unit that learns the correlation between the required time of each of the links and the round trip time based on the input data and the teacher signal, and based on the learning result, the test data specified by the test data. A traffic estimation device in a packet switching network, comprising: a neural network that outputs an expected transit time of an arbitrary route.