JP5901681B2 - Apparatus and method for estimating maximum flow rate in a short time - Google Patents

Description

  The present invention relates to a short-time maximum flow rate estimation apparatus and method, and more particularly to a short-time maximum flow rate estimation apparatus and method for measuring traffic fluctuations and estimating a maximum flow rate within a short time.

  In an existing ATM (Asynchronous Transfer Mode) network, traffic characteristics such as parameters determined in advance such as peak speed and maximum burst size, or parameters statistically determined using a probability model are used. And quality of service is determined (for example, see Non-Patent Document 1).

  In addition, in order to obtain the behavior of traffic flowing in the network, actual traffic measurement is performed at very short observation intervals, long-term fluctuation behavior of traffic is removed from the measured value, and short-term fluctuation is statistically predicted from its variance. Alternatively, there is a calculation method (Patent Document 1).

  However, in a network that has recently been increased in speed, it is very burdensome to measure a large amount of traffic with a fine granularity, and it is difficult to use it constantly. Therefore, in the traffic management technique, a method for easily obtaining the maximum flow rate of traffic flowing through the network is desired. As a technique in the ATM network shown in Non-Patent Document 1, there is a method for estimating a maximum flow rate using a traffic flow rate for each individual flow as a variable. This is an important technique (hereinafter referred to as “Prior Art 1”) for estimating how the maximum flow rate changes according to the traffic flow rate for each flow. However, in general, it is difficult to acquire all the traffic flow rates for each flow, and therefore it is possible to evaluate by assuming that the traffic flow rate of all members is uniform. However, the estimation accuracy of the maximum flow rate tends to decrease. In order to estimate the maximum flow rate under the condition that the traffic flow rate for each flow is not known, a method for solving the nonlinear programming problem (hereinafter referred to as “Prior Art 2”) has also been proposed (Patent Document 2). . As an element common to the prior art 1 and the prior art 2, there is a feature that an upper limit speed for each flow is required as a parameter.

JP 2004-247957 A JP 2009-206698 A

ATM Network, Masayuki Murata, Science and Technology Publishing, 1997

  However, if communication at a constant speed such as conventional voice calls is excluded, the communication speed changes with time in actual communication, so the peak speed is set for flows with a large proportion of time to communicate at peak speed. However, the peak speed is often a value that can only be realized in an ideal environment for the communication method, and the actual speed is lower than the peak speed and the fluctuation is smaller. However, the maximum flow rate tends to be excessively estimated as traffic with large short-term fluctuations.

  As described above, the maximum flow rate estimation method using the theoretical value of peak speed tends to estimate short-term fluctuations and estimate the maximum flow rate to a large extent, but it is difficult to measure actual traffic at short-time intervals. It is.

  The present invention has been made in view of the above points, and suppresses the load related to measurement as much as possible, and avoids overestimating short-term fluctuations in actual traffic, thereby reducing the maximum flow rate of traffic flowing through a communication network. It is an object of the present invention to provide an apparatus and method for estimating a maximum flow rate within a short time that can be estimated.

According to one aspect, a short-time maximum flow rate estimation device that estimates traffic flow rate in a communication network in consideration of short-term traffic fluctuations,
A flow traffic extracting means for collecting the flow traffic;
Flow traffic analysis means for analyzing the flow traffic and calculating statistics;
According to the statistic, a flow model estimation unit that classifies into one or more types of traffic models by parameter estimation and accumulates in a flow information accumulation unit;
A multiplicity calculation unit for each model for determining a flow multiplicity for each traffic model obtained by virtually multiplexing the flow represented by the traffic model accumulated in the flow information accumulation unit for each traffic model;
Multiple traffic model construction means for calculating a traffic model after multiplexing based on the flow multiplicity from the statistics and individual traffic models, and calculating a short-time variation characteristic of the traffic model after the multiplexing,
There is provided a short-time maximum flow rate estimation device comprising: a maximum flow rate calculation means for estimating the magnitude of the maximum flow rate from the short-time fluctuation characteristics.

  According to one aspect, when performing traffic measurement in a communication network, it does not measure in detail the entire traffic in which a large number of flows exist at the same time, but relates to traffic measurement by measuring only the traffic of individually extracted flows. The load is reduced. Furthermore, by evaluating the short-term fluctuation characteristics of the overall traffic based on the traffic model constructed based on the detailed information of the flow traffic, the shortness of the overall traffic evaluated based on the traffic model based on the peak speed information for each flow. It is possible to estimate the maximum flow rate with higher accuracy than using the time variation characteristic. Since the maximum flow rate is used as an extension determination criterion for communication links and communication devices, the smaller maximum flow rate can be correctly estimated, which has the effect of enabling more efficient design of communication equipment.

The system configuration example of this invention. The structural example of the maximum flow volume estimation apparatus in the 1st Embodiment of this invention. The flowchart of the process of the maximum flow volume estimation apparatus in the 1st Embodiment of this invention. The structural example of the maximum flow volume estimation apparatus in the 2nd Embodiment of this invention. The flowchart of the process of the maximum flow volume estimation apparatus in the 2nd Embodiment of this invention. Numerical values used in the examples of the present invention. The example of the maximum flow volume estimated value in the Example of this invention.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 shows a system configuration example of the present invention.

  In a communication network system to which the present invention is applied, a communication apparatus A102 and a communication apparatus B103 are connected via a communication link 101. In the example of FIG. 1, the maximum flow rate estimation apparatus 104 is connected to one communication apparatus A102. . The maximum flow rate estimation device 104 captures a measurement target by mirroring or the like, and acquires an average traffic flow rate from MIB (Management Information base) information.

[First Embodiment]
FIG. 2 shows a configuration example of the maximum flow rate estimation apparatus in the first embodiment of the present invention.

  The maximum flow rate estimation apparatus 104 shown in the figure includes a flow traffic extraction unit 201, a flow traffic analysis unit 202, a flow model estimation unit 203, a flow information accumulation DB 204, a model-specific multiplicity calculation unit 205, a multiple traffic model construction unit 206, a maximum A flow rate calculation unit 207 is included.

  The processing of the maximum flow rate estimation device 104 in the above configuration is shown below.

  FIG. 3 is a flowchart of the operation of the maximum flow rate estimation apparatus according to the first embodiment of the present invention.

  Step 101) The flow traffic extracting unit 201 randomly selects a specific flow from the connected communication device A102, and extracts data of only the traffic of the flow.

  Step 102) The flow traffic analysis unit 202 analyzes the traffic data of the flow extracted by the flow traffic extraction unit 201, and calculates statistics such as average, variance, and peak speed.

  Step 103) The flow model estimation unit 203 determines one or more appropriate traffic models for the flow based on the statistics calculated by the flow traffic analysis unit 202. As a traffic model, there are a method using a model expressed by three types of communication rates of 0, a peak value, and an intermediate value, and a method using a model that approximates a distribution of communication rates with a normal distribution.

  Step 104) The flow information accumulation DB 204 records flow information such as the traffic model and statistics, communication start and end times of the flow obtained by the flow model estimation unit 203.

  Step 105) The model-specific multiplicity calculation unit 205 reads all or a part of the flow information recorded in the flow information accumulation DB 204, and sets the average traffic flow acquired from the communication device A102 or a predetermined average traffic flow. The virtual flow multiplicity is determined for each traffic model recorded in the flow information accumulation DB 204 so as to match.

  Step 106) The multiple traffic model construction unit 206 calculates the traffic model after multiplexing the number of flows determined by the model-specific multiplicity calculation unit 205 based on the statistics and the traffic model of each flow, The short-time fluctuation characteristic of the calculated traffic model is calculated. Conventional techniques for grasping characteristics of short-term fluctuations are generally modeled by observing the behavior of the entire traffic after flow multiplexing, but in the present invention, the behavior of a single flow is measured. And the characteristics of the entire traffic are evaluated as those obtained by superimposing the modeled flows.

  Step 107) The maximum flow rate calculation unit 207 calculates, as the maximum flow rate, the maximum traffic flow generated by the multiple traffic or the upper limit value that can be suppressed to a certain probability or less from the short-time fluctuation characteristics calculated by the multiple traffic model construction unit 206.

[Second Embodiment]
In this embodiment, in addition to the processing of the first embodiment, an example in which the maximum flow multiplicity satisfying the maximum flow rate that does not exceed the capacity of the transmission path is obtained is shown.

  FIG. 4 shows an example of the configuration of the maximum flow rate estimation apparatus in the second embodiment. In this figure, the same parts as those in FIG. The configuration shown in FIG. 4 is a configuration in which a maximum flow rate determination unit 308 is added to the configuration of FIG.

  FIG. 5 is a flowchart of the process of the maximum flow rate estimation apparatus in the second embodiment of the present invention.

  Step 201) The flow traffic extraction unit 201 selects a specific flow from the connected communication device A102, and extracts data of only the traffic of the flow.

  Step 202) The flow traffic analysis unit 202 analyzes the traffic data of the flow extracted by the flow traffic extraction unit 201, and calculates statistics such as average, variance, and peak speed.

  Step 203) The flow model estimation unit 203 determines an appropriate traffic model for the flow based on the statistics calculated by the flow traffic analysis unit 202.

  Step 204) The flow information accumulation DB 204 records flow information such as the flow traffic model and statistics, communication start and end times obtained by the flow model estimation unit 203.

  Step 205) The model-specific multiplicity calculation unit 205 reads all or a part of the flow information recorded in the flow information accumulation DB 204, and sets the average traffic flow acquired from the communication device A102 or a predetermined average traffic flow. The flow multiplicity is determined for each traffic model recorded in the flow information accumulation DB 204 so as to match. The average traffic flow rate and flow multiplicity satisfy the maximum flow rate that does not exceed the capacity that can be accommodated in the transmission path.

  Step 206) The multi-traffic model construction unit 206 uses the traffic model after multiplexing the number of flows (flow multiplicity) determined by the multiplicity calculation unit 205 for each model as a statistic and the traffic model of each flow. And the short-time fluctuation characteristic of the calculated traffic model is calculated.

  Step 207) The maximum flow rate calculation unit 207 calculates, as the maximum flow rate, the maximum traffic flow generated by the multiple traffic or the upper limit value suppressed to a certain probability or less from the short-time fluctuation characteristics calculated by the multiple traffic model construction unit 206.

  Step 208) When the calculated maximum flow rate is smaller than the maximum allowable flow rate set for the communication link through which the traffic to be measured flows, the maximum flow rate determination unit 308 proceeds to Step 209. If the calculated maximum flow rate is equal to or greater than the maximum allowable flow rate set for the communication link through which the traffic to be measured flows, the process proceeds to step 210.

  Step 209) The model multiplicity calculation unit 205 again increases the flow multiplicity, and the process proceeds to Step 207, where the maximum flow rate calculation is repeated and used for the calculation just before the maximum flow rate calculation value exceeds the maximum allowable flow rate. Let the average traffic flow be the maximum accommodated traffic flow.

  Step 210) Again, the multiplicity calculation unit 205 for each model decreases the flow multiplicity, and the process proceeds to Step 107, where the maximum flow rate calculation is repeated and used when the maximum flow rate calculation value is below the maximum allowable flow rate for the first time. Let the average traffic flow be the maximum accommodated traffic flow.

  The above processing will be specifically described below.

An example of maximum flow rate estimation in a link that accommodates mobile high-speed communication, represented by a smart phone, is shown below. As shown in FIG. 6, assuming that the capacity of the link accommodating high-speed communication is 10 Gbps, the average traffic flow rate is 3 Gbps, and the average number of flows flowing through the link is 7,500, the average speed per flow is 0.4 Mbps. Assuming that the peak speed per flow of mobile communication is 112.5 Mbps, in the conventional technology example, each flow is regarded as an on / off model with a peak R1 = 112.5 Mbps and an average A1 = 0.4 Mbps, and short-term fluctuations in the overall traffic are evaluated. It was. Specifically, if the communication speed of the _Nth flow is expressed as XN,
P {X _N = peak speed} = p1 = A1 / P1 = 0.4 / 112.5 = 0.0036
P {X _N = 0} = p2 = 1-P {X _N = peak speed} = 0.964
And for a sufficiently small α (eg 10 ^-3 etc.)
P {X ₁ + X ₂ + ... + X _7,500 ≧ + BW1} <α
The minimum BW1 that satisfies the above has been obtained by direct calculation or approximation. For example, when approximated using a normal distribution, the distribution of communication speed per flow is
V1 = A1 × (R1-A1)
= 0.4 × (112.5-0.4)
= 44.84
And the distribution of the multiplexed traffic is
V1all = V1 x (total number of flows)
= 44.84 × 7,500
= 336,300
And the standard deviation is
S1 = V1all ^1/2 = 579.9 (Mbps)
The value of BW1 for α = 10 ^-3 is
BW1 = 0.4 × 7,500 + 3.09 × 579.9 = 4,791.891 (Mbps)
Is calculated.

Actually, it is rare that the peak speed continues to appear in the traffic per flow, and the variance of the communication speed of the flow is expected to be considerably smaller than V1. Now, by the individual flow measurement in the flow traffic analysis unit 202, the average flow rate A2, the variance V2, the probability q1 that the communication speed is greater than 0, and the probability that the communication speed of the flow is 0 are obtained by q2 (= 1−q1). Suppose. At this time, the average A2 ′ of communication speed when the communication speed of the flow is 0 or more, and the variance V2 ′ are
A2 '= A2 / q1, V2' = (V2 + A2 ² ) / q1-A2 ' ²
Is required. A model is assumed in which the peak speed R1 and another speed R2 are taken as values when the flow communication speed is 0 or more. When the probability that the communication speed of the flow is R1 is r1, and the probability that it is R2 is r2 (= q1-r1), R2, r1, and r2 are
(r1 / q1) R1 + (r2 / q1) R2 = A2 ',
(r1 / q1) R1 ² + (r2 / q1) R2 ² = V2 '+ A2' ²
It can be calculated as a solution of Assume that the average flow rate A2 = 0.4 Mbps, the variance V2 = 17.66, the peak rate R1 = 112.5 Mbps, and the probability that the communication rate is greater than 0 q1 = 0.021. At this time,
A2 '= 0.4 / 0.021 = 19.04Mbps , V2' = (17.66 + 0.4 2) /0.021-19.04 2 = 486.05
R2, r1, and r2 are the solutions of the simultaneous equations,
R2 = 13.84Mbps, r1 = 0.00111, r2 = 0.0199
Is obtained. Therefore, the flow model estimation unit 203 is modeled as a flow communicating at a rate of 0 with probability 1-q1 = 0.979, a rate of R2 = 13.84 Mbps with a probability of 0.0199, and a rate of R1 = 112.5 Mbps with a probability of 0.00111. Because this flow can suppress fluctuations for a shorter time than the conventional on-off model,
P {X ₁ + X ₂ + ... + X _7,500 ≧ + BW2} <α
The value of BW2 that satisfies the condition can be estimated smaller than BW1 obtained by the existing method. When approximated using a normal distribution, the variance of the multiplexed traffic is
V2all = V2 x (total number of flows)
= 17.66 x 7,500
= 132,450
Standard deviation is
S2 = V2all ^1/2 = 363.9 (Mbps)
The value of BW2 for α = 10 ^-3 is
BW2 = 0.4 x 7,500 + 3.09 x 363.9 = 4,124.451 (Mbps)
Is calculated. As a result, the maximum flow rate estimate BW2 could be suppressed to 86% of BW1.

  FIG. 7 shows the maximum flow rate estimated value based on the conventional on / off model and the maximum flow rate estimated value based on the flow measurement of the present invention.

  Furthermore, since higher-order moments above the average and variance cannot be used with existing methods, models that can be approximated are limited, but with the method of the present invention, third-order or higher moments can also be used, so a wider range of probability models can be used. It is also possible to increase the approximation accuracy.

  The operation of the constituent elements of the maximum flow rate estimation device 104 shown in FIGS. 2 and 4 is constructed as a program and installed in a computer used as the maximum flow rate estimation device, or distributed via a network. Is possible.

  The present invention is not limited to the above-described embodiments, and various modifications and applications are possible within the scope of the claims.

101 communication link 102 communication device A
103 Communication device B
104 Maximum flow rate estimation device 201 Flow traffic extraction unit 202 Flow traffic analysis unit 203 Flow model estimation unit 204 Flow information storage DB
205 Multiplicity Calculation Unit by Model 206 Multiple Traffic Model Construction Unit 207 Maximum Flow Rate Calculation Unit 308 Maximum Flow Rate Determination Unit

Claims (8)

A short-time maximum flow rate estimation device that estimates traffic flow rate in a communication network in consideration of short-term traffic fluctuations,
A flow traffic extracting means for collecting the flow traffic;
Flow traffic analysis means for analyzing the flow traffic and calculating statistics;
According to the statistic, a flow model estimation unit that classifies into one or more types of traffic models by parameter estimation and accumulates in a flow information accumulation unit;
A multiplicity calculation unit for each model for determining a flow multiplicity for each traffic model obtained by virtually multiplexing the flow represented by the traffic model accumulated in the flow information accumulation unit for each traffic model;
Multiple traffic model construction means for calculating a traffic model after multiplexing based on the flow multiplicity from the statistics and individual traffic models, and calculating a short-time variation characteristic of the traffic model after the multiplexing,
Maximum flow rate calculating means for estimating the maximum flow rate from the short-time fluctuation characteristics;
A short-term maximum flow rate estimation device characterized by comprising: The model-specific multiplicity calculating means includes:
2. The short time according to claim 1, further comprising means for determining the flow multiplicity for each traffic model recorded in the flow information storage means so that the flow multiplicity for each model matches a predetermined average traffic flow rate. Maximum flow rate estimation device. The maximum flow rate calculating means includes
2. The method according to claim 1, further comprising means for obtaining a maximum value of the average traffic flow rate that can be accommodated in the transmission line by obtaining a maximum flow multiplicity and an average traffic flow rate that satisfy a maximum flow rate that does not exceed the capacity of the transmission line. Maximum flow rate estimation device. The flow model estimation means includes:
2. The short-time maximum flow rate estimation device according to claim 1, wherein a model represented by three types of communication rates of 0, a peak value, and an intermediate value is used as a traffic model. The flow model estimation means includes:
2. The short-time maximum flow rate estimating apparatus according to claim 1, wherein the traffic rate model approximates a communication rate distribution by a normal distribution. A method for estimating a maximum flow rate in a short time in a device for estimating a traffic flow rate in a communication network in consideration of short-term traffic fluctuations,
A flow traffic extraction step for collecting flow traffic; and
A flow traffic analysis step of analyzing the flow traffic and calculating a statistic;
According to the statistics, a flow model estimation step for classifying into one or more types of traffic models by parameter estimation and storing in a flow information storage unit;
A model-specific multiplicity calculation step for determining a flow multiplicity for each traffic model obtained by virtually multiplexing the flow represented by the traffic model stored in the flow information storage means, for each traffic model;
A multi-traffic model construction step of calculating a traffic model after multiplexing based on the flow multiplicity from the statistics and individual traffic models, and calculating a short-time variation characteristic of the traffic model after the multiplexing,
A maximum flow rate calculating step for estimating the maximum flow rate from the short-time fluctuation characteristics;
A method for estimating a maximum flow rate in a short time, characterized by: In the model-specific multiplicity calculation step,
7. The maximum flow rate in a short time according to claim 6, wherein the flow multiplicity for each traffic model recorded in the flow information storage means is determined so that the flow multiplicity for each model matches a predetermined average traffic flow rate. Estimation method. In the maximum flow rate calculating step,
The maximum flow rate estimation within a short time according to claim 6, wherein the maximum value of the average traffic flow rate that can be accommodated in the transmission line is obtained by obtaining the maximum flow multiplicity and the average traffic flow rate that satisfy the maximum flow rate not exceeding the capacity of the transmission line. Method.

Images