JP2009118272A - Program for network band estimation, apparatus for network band estimation, method for network band estimation, and measuring apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To estimate network load without applying load to a network. <P>SOLUTION: An apparatus 1 for network band estimation receives through the network a test packet which a test packet transmitting terminal 2 has inserted a timestamp into and transmitted, estimates the loaded condition of the network from the state of the test packet delay. On the occasion, when the background traffic of the network is a sound traffic, the relational expression of the network load and the test packet delay is found based on an M/D/1 queueing model, when the background traffic of the network is a video image traffic, the relational expression of the network load and the test packet delay is found based on an M/M/1 queueing model, or an M/G/1-processor sharing queueing model. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a network bandwidth estimation technique for estimating a network bandwidth, and more particularly to a bandwidth estimation technique based on test packet behavior.

  In order to guarantee network communication quality, it is necessary to manage the amount (bandwidth) of network resources. Although there is a packet capture tool (device) as a device for measuring the network bandwidth, it lacks scalability because the cost of the device itself is high and all packets need to be captured.

  As load measurement tools, many software that send test packets to a network and estimate the network bandwidth on the measurement path based on delay information of the test packets have been proposed. These tools are methods for estimating the available bandwidth by changing the transmission pattern of the test packet and searching for points where the packet reception behavior changes.

  For example, in Patent Document 1, a plurality of periodic time stamp packets are transmitted to a receiving communication device at a currently set test transmission rate (R), and each time stamp packet is transmitted to the receiving communication device. Based on the received time, the change tendency of the transmission delay time difference between the time stamp packets is checked. If the transmission delay time difference is not included in the predetermined stable range and shows an increasing tendency, the inspection is repeated after reducing R to a value. On the other hand, if the transmission delay time difference is not included in the stable range and shows a decreasing tendency, the inspection is repeated with the value increased by R. As a result of such processing, when the transmission delay time difference is included in the stable range, the currently set test transmission rate is determined as the available bandwidth.

Further, in Patent Document 2, continuous transmission of probe packets is repeated while changing the transmission interval I _k of probe packets from the transmission terminal to the reception terminal, and the reception terminal sequentially measures the reception intervals I _k ′ of the received probe packets. To do. Then, the available bandwidth of the communication path between the transmitting terminal and the receiving terminal is calculated by linearly approximating the obtained series of data (I _k , I _k ′ −I _k ). Note that, as a linear approximation, a linear function y = −ax + b is used, and the available bandwidth B is obtained by the equation B = (a × L) / b.

JP 2006-074773 A Japanese Patent Laid-Open No. 2004-254025

  The conventional bandwidth estimation method described above is a method for estimating the available bandwidth by gradually changing the test traffic and determining the change point of the test packet delay characteristic. Therefore, there is a possibility that the bandwidth is temporarily occupied as test traffic every time the available bandwidth is estimated, and the bandwidth cannot be used effectively.

  In addition, since the conventional technology does not consider the packet delay that occurs depending on the devices constituting the network, the network load cannot be accurately estimated, and the load is estimated to be larger than the actual value. There was a problem that there was a possibility.

  The present invention has been made to solve the above-described problems in the prior art and to solve the problems. The present invention estimates the network load without imposing a burden on the network, and takes into account delays caused by network components. It is an object of the present invention to provide a network bandwidth estimation program, a network bandwidth estimation device, a network bandwidth estimation method, and a measurement device that can perform highly accurate bandwidth estimation.

  In order to solve the above-described problems and achieve the object, the present invention selects a traffic model based on whether the network traffic type is video traffic or voice traffic, and information about the network traffic and the traffic A relational expression indicating the relationship between the network load and the delay is obtained from the model, a test packet is received from the network, and the use band of the network is estimated based on the delay state of the test packet and the relational expression.

  In the present invention, when the traffic type is voice traffic, the M / D / 1 queue model is selected as the traffic model, and when the traffic type is video traffic, the traffic model is M / M. Select the / 1 queue model or the M / G / 1-Processor Sharing queue model to find the relational expression.

を前記関係式として導出し、前記トラフィック種別が映像トラフィックである場合にはＭ/Ｍ/１待ち行列モデル、あるいはM/G/1-Processor Sharing待ち行列モデルから

を前記関係式として導出する。 In addition, the present invention obtains the average time μ required for processing one packet from information related to network traffic, and the traffic type is voice for the relationship between the average value ave [fluc] of the delay fluctuation amount of the test packet and the network load ρ. If it is traffic, from the M / D / 1 queuing model

Is derived as the relational expression, and when the traffic type is video traffic, it is derived from the M / M / 1 queue model or the M / G / 1-Processor Sharing queue model.

Is derived as the relational expression.

を前記関係式として導出し、前記トラフィック種別が映像トラフィックである場合にはＭ/Ｍ/１待ち行列モデルから

を、あるいはM/G/1-Processor Sharing待ち行列モデルから

を前記関係式として導出する。 In the present invention, the time μ required for processing one packet is obtained from information relating to network traffic, and the relationship between the dispersion value Var of the delay fluctuation amount of the test packet and the network load ρ is a voice traffic type. From the M / D / 1 queuing model

From the M / M / 1 queue model when the traffic type is video traffic.

Or from the M / G / 1-Processor Sharing queuing model

Is derived as the relational expression.

  In addition, the present invention derives a relational expression using the average packet size of the target traffic and the link physical bandwidth information to be monitored as information relating to network traffic.

  Further, the present invention analyzes the delay state of the test packet when the network is unloaded, and corrects the evaluation by the delay state evaluation step.

  In the present invention, the traffic type of the network is discriminated, and the traffic model is selected based on the discrimination result.

  According to the present invention, the “network load and delay relation model” corresponding to the traffic type is automatically constructed, and the network bandwidth estimation is realized by referring to the relation model with the measurement delay in the test traffic. The effect of being able to obtain a network bandwidth estimation program, a network bandwidth estimation device, a network bandwidth estimation method, and a measurement device that enable highly accurate bandwidth estimation for each traffic type and enable bandwidth estimation with minute test traffic Play.

  Further, according to the present invention, it is possible to evaluate by removing a delay element depending on a transmission / reception terminal from the measured packet delay, and a network bandwidth estimation program, a network bandwidth estimation device, which realizes high accuracy of bandwidth estimation, There is an effect that a network bandwidth estimation method and a measurement device can be obtained.

  Further, according to the present invention, by setting the type of traffic to be measured, the average packet size of the target traffic, and the link physical bandwidth information to be monitored as input parameters, the “relationship between network load and delay in the measurement environment” There is an effect that a network bandwidth estimation program, a network bandwidth estimation device, a network bandwidth estimation method, and a measurement device that make it possible to automatically construct a “model” can be obtained.

  Further, according to the present invention, by setting a time interval for calculating an average value or variance value of measurement delay fluctuations, it is possible to calculate a network average use bandwidth in the time interval, and a designation including a minute time. It becomes possible to monitor the band average value in the time interval.

  In addition, according to the present invention, in a single traffic network of voice traffic or video traffic, it is possible to perform highly accurate bandwidth estimation reflecting traffic characteristics by switching the bandwidth estimation model for each traffic type. Become.

  Embodiments of a network bandwidth estimation program, a network bandwidth estimation device, a network bandwidth estimation method, and a measurement device according to the present invention will be described below in detail with reference to the drawings.

  FIG. 1 is a schematic configuration diagram illustrating a schematic configuration of a network bandwidth estimation apparatus according to an embodiment of the present invention. As shown in the figure, the network bandwidth estimation device 1 is connected to the test packet transmission terminal 2 via a network composed of nodes N1, N2, and the like.

  The test packet transmission terminal 2 has a timer 20 and a packet transmission unit 21 therein. The packet transmission unit 21 is a functional unit that transmits a test packet to the network, and inserts a time stamp into the test packet with reference to the timer 20.

  The network bandwidth estimation device 1 includes a timer 10, a packet receiving unit 11, a delay measuring unit 12, an initial delay analyzing unit 13, a bias calculating unit 14, a delay fluctuation calculating unit 15, a parameter setting unit 16, and a traffic type determining unit 17 therein. A model construction unit 18 and a band estimation unit 19.

  The packet receiving unit 11 is a functional unit that receives a packet from the network. Specifically, the packet receiving unit 11 captures a test packet transmitted by the test packet transmitting terminal 2 and a background traffic packet to be monitored flowing on the network.

  The timer 10 is synchronized with the timer 20 of the test packet transmission terminal 2. The delay measuring unit 12 calculates a difference value between the packet time stamp information and the packet reception time obtained by the timer 10.

  The initial delay analysis unit 13 determines a minimum value for the time difference value in the delay measurement unit 12 in a certain period at the time of initial measurement. Based on the output of the initial delay analysis unit 13, the bias calculation unit 14 preliminarily measures a delay fluctuation amount (bias amount) specific to the terminals constituting the network.

  The delay fluctuation calculation unit 15 removes the terminal-specific influence derived in the bias calculation unit from the relative variation from the minimum value in the initial delay analysis unit with respect to the time difference value of the test packet in the delay measurement unit 12. Is calculated.

  The parameter setting unit 16 provides the model construction unit 18 with information on the average packet size of the monitoring target traffic and the link physical bandwidth of the measurement path necessary for constructing the bandwidth estimation model. This parameter may be input by an operator, or may be identified from background traffic received by the packet receiver 11.

  The traffic type determination unit 17 determines the type of traffic to be measured based on operator settings or estimates it from the measured packet characteristics and provides it to the model construction unit 18.

  The model construction unit 18 selects a predefined traffic model from the traffic type determined by the traffic type determination unit 17 and derives a “relational expression between network load and delay” based on the parameters obtained from the parameter setting unit 16. To do.

  The bandwidth estimation unit 19 estimates the traffic network usage band by referring to the “relational expression of network load and delay” generated by the model construction unit 18 with the delay fluctuation amount acquired by the delay fluctuation calculation unit 15. .

  That is, in the network bandwidth estimation method according to the present invention, as shown in FIG. 2, a test packet is transmitted by a network measurement system connected to a specific relay node pair and having both packet transmission and packet reception / analysis functions. Based on the delay behavior of the received packet, the bidirectional network bandwidth is estimated.

  At that time, a different model is selected based on whether the background traffic type of the network is video traffic or voice traffic, and a relational expression of load and delay is derived. As shown in FIG. Since the load is obtained on the basis of the load, it is possible to estimate the bandwidth state without imposing a burden on the network with a small amount of test traffic.

A basic processing flow of the network bandwidth estimation apparatus 1 is shown in FIG. As shown in the figure, the network bandwidth estimation apparatus 1 executes the following steps in the basic process.
(STEP A-1)
Enter the traffic type, average packet size of the target traffic, and link physical bandwidth information to be monitored.
(STEP A-2)
Established “Relationship between network load and delay” for monitored traffic.
(STEP B-1)
Receive test packet.
(STEP B-2)
The average value / variance value of the delay fluctuation in the test packet is calculated.
(STEP 3)
Based on the statistical value (average / variance) obtained in (STEP B-2), by referring to the “relative expression model of network load and delay” corresponding to the target traffic derived in (STEP A-2) Estimate the bandwidth used for the monitored link.

Hereinafter, specific examples of each step will be described in detail.
(STEP A-1) Parameter Setting In this step, the traffic type of the target traffic, the average packet size of the target traffic, and the link physical bandwidth information to be monitored are set. As already described, the setting may be input by an operator or may be automatically set.

  When parameters are automatically set, the average packet size can be measured by capturing the target traffic. The traffic type can be determined by analyzing the traffic characteristics of each traffic (voice / video).

  FIG. 5 shows packet transmission patterns and packet sizes in the video traffic flow and audio traffic flow. As shown in the figure, analysis is performed for each single flow by destination / source address and port number, and traffic with a short packet size and weak burstiness can be regarded as a voice packet. For example, in a voice flow based on G.711, the payload size is 160 bytes, and the IP + UDP + RTP header is 40 bytes, so the total packet size is 200 bytes. Similarly, since the voice flow based on G.729 has a payload size of 20 bytes and an IP + UDP + RTP header of 40 bytes, the total packet is 60 bytes.

  As for the transmission frequency, for example, for a video flow of 6 Mbps and 30 frames / sec, it is common to packetize one frame (about 200 kbits) every 33 milliseconds and send it at once. If there are, a packet of about 160 kbytes is sent out every 20 milliseconds. Therefore, when the packet size is large (about 1200 to 1500 bytes) and the burstiness is strong, it can be regarded as a video flow, and when the packet size is small and the burstiness is weak, it can be regarded as an audio flow.

  The link physical band can be automatically determined by collecting information from relay nodes on the route using SNMP / MIB or the like.

(STEP A-2) Construction of relationship model between network load and delay Figure 6 shows the construction flow of the relationship model between network load and delay.
The parameter obtained in (STEP A-1) is the average packet size DS (bits)
Monitoring link speed C (bits / sec)
The time parameter μ for processing one packet is
μ = DS / C
As required.

  Next, when voice traffic is selected as the traffic type, a relational expression between the network load and the delay state is derived based on the queuing model M / D / 1.

となる。ここから、

を求めることができる。 When the average value ave [fluc] of the delay fluctuation amount is used as the delay state, the network load (ratio of the network use band to the physical band C) ρ: 0 to 1 is based on the following M / D / 1 mathematical model

It becomes. from here,

Can be requested.

ここで、Var_ref［ρ］は、ρを0〜1において任意刻み幅で値を与えたときの参照データである。このネットワーク負荷ρと分散参照値Var_refを保持しておく。 Further, when estimating the network load ρ: 0 to 1 using the dispersion value Var of the delay fluctuation amount as the delay state, first, the construction of the M / D / 1 reference model is constructed.

Here, Var _ref [ρ] is reference data when a value is given by an arbitrary step size in the range of ρ from 0 to 1. The network load ρ and the distributed reference value Var_ref are held.

  Then, as shown in FIG. 7, the measurement statistical value Var obtained by measuring the test packet is compared with the reference value Var_ref, and by selecting ρ corresponding to the reference value Var_ref closest to the measurement value, The network load ρ can be estimated.

  Next, when video traffic is selected, the traffic use band is estimated based on the queuing model M / M / 1 or M / G / 1-PS.

となる。ここから、

で、ネットワーク負荷を求めることができる。 First, when calculating the network load using the average value ave [fluc] of the delay fluctuation amount as the delay state, the M / M / 1 (= M / G / 1-PS) mathematical model is

It becomes. from here,

Thus, the network load can be obtained.

あるいはM/G/1-PSから

と表すことができる。ここで、Var_ref［ρ］は、ρを0〜1において任意刻み幅で値を与えたときの参照データである。このトラフィック負荷ρと分散参照値Var_refを保持して。おくことで、計測統計値であるVarと参照値であるVar_refを比較し、計測値と最も近い参照値Var_refに対応するρを推定負荷とする。 In addition, when estimating the network load ρ: 0 to 1 using the variance value Var of the delay fluctuation amount as the delay state, the M / M / 1 model is used.

Or from M / G / 1-PS

It can be expressed as. Here, Var _ref [ρ] is reference data when a value is given by an arbitrary step size in the range of ρ from 0 to 1. Hold this traffic load ρ and distributed reference value Var_ref. By comparing, the measurement statistical value Var and the reference value Var_ref are compared, and ρ corresponding to the reference value Var_ref closest to the measurement value is set as the estimated load.

(STEP B-1) Sending test packet As shown in Fig. 8, the test packet sending terminal 2, which is the network measuring device on the sending side, generates a test flow for measurement (for example, packet size 200 bytes sending interval 20msec) on the network. Then, the network bandwidth estimation device 1 which is the network measuring device on the receiving side receives the packet forming this test flow. The packet transmission interval may be random.

(STEP B-2) Statistical Calculation of Delay Fluctuation FIG. 9 shows a statistical calculation flow of delay fluctuation. The clocks between the two network measurement devices (the test packet transmission terminal 2 and the network bandwidth estimation device 1) that perform measurement are synchronized according to the prior art.

First, at the time of initial measurement of a test packet, as shown in FIG. 10, a “transmission / reception time difference” is measured for a certain time, and a “minimum delay” is calculated.
Minimum delay = MIN (di-si) i = 0,1, ..., N
Delay fluctuation fluc _i = (di−si) −minimum delay i = 0,1,…, N
Here, si is a time stamp at the time of packet transmission, and di is a packet arrival time.

として算出する。また、端末に依存する分散ゆらぎ成分は、

として算出する。 Next, in a measurement environment as shown in FIG. 11, that is, in a state where the network measurement devices are cross-connected or in a state where there is no other traffic, a delay variation amount depending on the measurement device is calculated. calculate. The average fluctuation component depending on the terminal is

Calculate as Also, the dispersion fluctuation component depending on the terminal is

Calculate as

として算出し、分散遅延ゆらぎは

として算出する。なお、上記統計値算出における時間間隔は、監視したい平均帯域の時間粒度に応じて任意に設定が可能である。 For delay fluctuation statistics,
The average delay fluctuation is

The dispersion delay fluctuation is calculated as

Calculate as The time interval for calculating the statistical value can be arbitrarily set according to the time granularity of the average bandwidth to be monitored.

  Next, a specific example of load estimation by the network bandwidth estimation apparatus according to the present embodiment will be described. The effect of the invention method was verified in the environment shown in FIG. In the verification network, network switches were installed on each switch to monitor the traffic load on the link between the two switches.

  Also, in the verification scenario, traffic of “1. VoIP only (1 Mbps aggregated flow)” or “2. Video only (9 Mbps single flow)” as the background traffic shall be stacked in order. The traffic load estimated by the method of the present invention was verified.

  In addition, 1. 1. In the VoIP-only verification scenario, all line bandwidths are 10 Mbps. In the video-only verification scenario, it was set to 100 Mbps, and the test flow sent short packet traffic that simulated 85.6 kbps VoIP traffic. The band estimation result is estimated at an average / dispersion calculation interval of about several tens of seconds.

  Each of FIGS. 13 to 16 shows an estimation result when the terminal-dependent delay fluctuation correction (delay bias correction) is not performed and an estimation result when the delay fluctuation correction is performed.

  When the background traffic load and the estimated load coincide with each other, the graph is plotted near the hatched line on the graph, and the ± 10% error line is also described in parallel with the hatched line. Also, compare the estimated results other than the “Relationship model between network load and delay” selected for each traffic type.

  As an effect of the delay fluctuation correction, the effect of the delay fluctuation correction is high when the band estimation is performed with the average value of the delay fluctuation as shown in FIGS.

  As an effect of model separation by traffic type, bandwidth estimation can be realized with an error of about ± 10% as an overall trend. In particular, the effect is great when band estimation is performed with a dispersion value of delay fluctuation as shown in FIGS.

  As described above, the model to be used is selected based on whether the network to be monitored is a video network or an audio network, and the relation between load and delay is derived, and the network is determined from the delay of the test packet. By estimating the load, it is possible to estimate the network load with a small amount of test traffic with high accuracy.

  In particular, when the network itself is used for a specific application such as audio or video, the load can be obtained with high accuracy by selecting a model.

  At this time, the load estimation accuracy can be improved by correcting the delay state of the test packet based on the initial delay state obtained in advance when the network is not loaded.

  The configuration and operation shown in this embodiment are merely examples, and the present invention is not limited to this. For example, the network band estimation apparatus shown in the present embodiment can be realized by software with some or all of the components. In particular, a network bandwidth estimation program that can obtain all of the components as software can be operated on any network device.

  As described above, the present invention is useful for network bandwidth estimation, and is particularly suitable for bandwidth estimation that does not impose a load on a network through which a specific type of traffic flows.

It is a schematic block diagram explaining the schematic structure of the network band estimation apparatus concerning a present Example. It is explanatory drawing explaining the network band estimation concerning a present Example. It is explanatory drawing explaining the process of a band estimation. It is a flowchart which shows the basic processing flow of the network band estimation apparatus shown in FIG. It is explanatory drawing explaining the traffic pattern in each traffic classification (video / audio). It is a construction flowchart of a relation model of network load and delay. It is explanatory drawing explaining the reference of a relationship model. It is explanatory drawing explaining transmission of a packet. It is explanatory drawing explaining the statistical value calculation of a delay fluctuation. It is explanatory drawing explaining the difference calculation of transmission / reception time. It is explanatory drawing explaining the measurement environment of delay bias amount. It is explanatory drawing explaining the specific example of the load estimation by the network band estimation apparatus concerning a present Example. It is an evaluation example of load estimation of voice traffic due to average fluctuation. It is an evaluation example of load estimation of voice traffic by distributed fluctuation. It is an evaluation example of the load estimation of the video traffic by the average fluctuation. It is an example of evaluation of the load estimation of the video traffic by dispersion | distribution fluctuation.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Network band estimation apparatus 2 Test packet transmission terminal 10,20 Timer 11 Packet receiving part 12 Delay measurement part 13 Initial delay analysis part 14 Bias calculation part 15 Delay fluctuation calculation part 16 Parameter setting part 17 Traffic type determination part 18 Model construction part 19 Bandwidth estimation unit 21 Packet transmission unit

Claims (10)

A network bandwidth estimation program for estimating a network bandwidth,
A model selection step of selecting a traffic model based on whether the traffic type of the network is video traffic or voice traffic;
A relational expression derivation step for obtaining a relational expression indicating a relation between network load and delay from the information relating to the traffic of the network and the traffic model;
Receiving a test packet from the network; and
A delay state evaluation step for evaluating a delay state of the test packet;
A bandwidth estimation step of estimating a use bandwidth of the network based on the delay state and the relational expression;
A network bandwidth estimation program characterized by causing a computer to execute.   In the model selection step, when the traffic type is voice traffic, an M / D / 1 queue model is selected as the traffic model, and when the traffic type is video traffic, M / D is selected as the traffic model. 2. The network bandwidth estimation program according to claim 1, wherein an M / 1 queue model or an M / G / 1-Processor Sharing queue model is selected. In the relational expression derivation step, the average time μ required for processing one packet is obtained from the information related to the traffic of the network, and the relationship between the average value ave [fluc] of the delay fluctuation amount of the test packet and the network load ρ,
From the M / D / 1 queuing model when the traffic type is voice traffic

Is derived as the relational expression,
If the traffic type is video traffic, use M / M / 1 queue model or M / G / 1-Processor Sharing queue model

The network bandwidth estimation program according to claim 2, wherein: is derived as the relational expression. In the relational expression derivation step, an average time μ required for processing one packet is obtained from information related to the traffic of the network, and the relationship between the dispersion value Var of the delay fluctuation amount of the test packet and the network load ρ,
From the M / D / 1 queuing model when the traffic type is voice traffic

Is derived as the relational expression,
If the traffic type is video traffic, from the M / M / 1 queue model

Or from the M / G / 1-Processor Sharing queuing model

The network bandwidth estimation program according to claim 2, wherein: is derived as the relational expression.   5. The relational expression deriving step derives the relational expression by using an average packet size of target traffic and link physical bandwidth information to be monitored as information related to traffic on the network. The network bandwidth estimation program according to claim 1.   An initial delay analyzing step for analyzing a delay state of the test packet when the network is in an unloaded state, and delay evaluation correcting means for correcting the evaluation by the delay state evaluating step based on an analysis result by the initial delay analyzing step. The network bandwidth estimation program according to any one of claims 1 to 5, wherein the network bandwidth estimation program is executed by a computer.   The computer further executes a type determining step for determining the traffic type of the network, and the model selecting step selects the traffic model based on a determination result of the type determining step. 5. The network bandwidth estimation program according to any one of 4 above. A network bandwidth estimation device for estimating a network bandwidth,
Model selection means for selecting a traffic model based on whether the traffic type of the network is video traffic or voice traffic;
Relational expression deriving means for obtaining a relational expression indicating a relation between network load and delay from the information regarding the traffic of the network and the traffic model;
Test packet receiving means for receiving a test packet from the network;
A delay state evaluating means for evaluating a delay state of the test packet;
Band estimation means for estimating a use band of the network based on the delay state and the relational expression;
A network bandwidth estimation apparatus comprising: A network bandwidth estimation method for estimating a network bandwidth, comprising:
A model selection step of selecting a traffic model based on whether the network traffic type is video traffic or voice traffic;
A relational expression derivation step for obtaining a relational expression indicating a relation between network load and delay from the information relating to the traffic of the network and the traffic model;
A test packet receiving step of receiving a test packet from the network;
A delay state evaluation step for evaluating a delay state of the test packet;
A bandwidth estimation step of estimating a bandwidth of use of the network based on the delay state and the relational expression;
A network bandwidth estimation method comprising: A measuring device having a function of estimating a network bandwidth,
Model selection means for selecting a traffic model corresponding to the traffic type to be measured,
Relational expression deriving means for obtaining a relational expression indicating a relation between network load and delay from the information relating to the traffic of the network and the traffic model selected from the model selecting means;
Delay state evaluating means for evaluating a delay state of a test packet received from the network;
A measurement apparatus comprising: band estimation means for estimating a use band of the network based on the delay state and the relational expression.

Images