JP4097606B2 - Communication band design management method and communication band design management apparatus - Google Patents

Description

  The present invention relates to a communication band design management method and a communication band design management apparatus, and more particularly to a technology effective when applied to link design and communication quality management through which different classes of user flows in an IP network pass.

  Conventionally, as IP networks are widely used, there is an increasing demand for guaranteeing communication quality over IP networks. On the other hand, it is necessary to reduce the cost for communication facilities. Therefore, a communication band design management method that appropriately calculates a communication band necessary for maintaining desired communication quality and manages whether the current communication quality can be maintained has become important.

  In the current IP network management, whether or not a certain link is congested is determined by comparing a predetermined threshold for the link usage rate with the measured link usage rate (for example, non-patent document). 1, see Non-Patent Document 2 and Non-Patent Document 3.) However, since the relationship between the link usage rate and the quality such as the file transfer time and throughput experienced by the user is unknown, it is difficult to specify the threshold for the usage rate that is judged to be a congestion state. .

  In addition, as a method of calculating the necessary bandwidth, there is conventionally a method of setting an IP level packet loss rate, or an ATM level cell loss rate or queuing delay time to a certain target value or less (for example, non- (See Patent Literature 4, Non-Patent Literature 5, Non-Patent Literature 6, Patent Literature 1, Patent Literature 2, Patent Literature 3, Patent Literature 4, Patent Literature 5). However, in these methods, the relationship between the IP level quality such as the packet loss rate and delay time and the TCP level quality such as the file transfer time and throughput experienced by the user is unclear, so it is difficult to set the target value. There was a problem that there was.

  Thus, as a method for solving such a problem, a method in consideration of TCP level quality has recently been proposed (for example, Non-Patent Document 7, Non-Patent Document 8, Non-Patent Document 9, Non-Patent Document 10, (See Non-Patent Document 11 and Non-Patent Document 12.)

However, in the methods described in Non-Patent Document 7, Non-Patent Document 8, Non-Patent Document 9, Non-Patent Document 10, Non-Patent Document 11, Non-Patent Document 12, etc., all the flows multiplexed on the link of interest. It requires the assumption that the round trip time and access link speed are uniform. For this reason, there is a problem that it is not possible to cope with a case where flows having different access line bandwidths or flows having different round trip times are multiplexed. As a method for solving this problem, for example, an average value r ′ of file transfer speeds of all flows is estimated from a set of the number of flows and link usage rates that can be easily measured (r ′ is r ′ = s / T is defined _ideal. here, s is the average file size, average file transfer time of the entire flow when T _ideal is small enough utilized link aggregating flows), link utilization is increased as a parameter A method of predicting an average file transfer time when ρ is reached and designing a necessary bandwidth has been proposed (for example, see Non-Patent Document 13).

However, the method described in Non-Patent Document 13 can handle only the average file transfer time of the entire flow. For this reason, for example, there is a problem that it is not possible to set a service class for each service contract type of a user such as a fee and manage and design a user flow by class. In recent years, a technique for differentiating services between classes such as Diffserv (Differentiated services) has attracted attention, and it is important to manage quality for each contracted class. Further, the required communication quality may differ depending on the application type. For example, if you download a small file like the web and display it on your terminal and access the next page, the request for response time will be strict, but download a large file like ftp In this case, it is assumed that the request for the download time becomes moderate. In such a case, it is necessary to design the bandwidth and manage the communication quality in response to different quality requirements for each class. However, the methods proposed so far cannot solve such a demand. was there.
JP 2000-4232 A JP-A-11-341000 JP-A-9-162867 Japanese Patent Laid-Open No. 11-154952 JP 2001-53804 A N. Ansari, G. Cheng, S. Israel, Y. Luo, J. Ma, and L. Zhu, "QoS provision with path protection for next generation SONET," ICC2002, 2002. A. Feldmann, A. Greenberg, C. Lund, N. Reingold, and J. Rexford, "Netscope: Traffic engineering for IP networks," IEEE Network, pp. 11-19, March / Apl. 2000. A. Kumar, M. Hegde, SVR Anand, BN Bindu, D. Thirumurthy, and AA Kherani, "Nonintrusive TCP connection admission control for bandwidth management of an Internet access link," IEEE Commun. Mag., Pp. 160-167, May 2000. Y. Kamado and K. Miyake, "A measured-traffic-based bandwidth dimensioning method for Internet ATM backbone networks," IEICE Trans. Commun., Vol. E81-B, no. 2, Feb. 1998. J.-s. Li, "Measurement and in-service monitoring for QoS violations and spare capacity estimations in ATM network," Computer Communications, 23, pp. 162-170, 2000. Yokoi, Tsuchiya, "VP capacity design method based on estimation of cell number distribution in buffer," IEICE Technical Report, SSE99-16, pp. 25-29, 1999. SB Fredj, T. Bonald, A. Proutiere, G. Regnie, and JW Roberts, "Statistical bandwidth sharing: A study of congestion at flow level," ACM SIGCOMM 2001, 2001. Ozawa, "Approximate Analysis of Communication Response Time Using Processor Sharing Model," IEICE Technical Report, NS2002-12, pp. 45-48, 2002. JW Roberts and L. Massoule, "Bandwidth sharing and admission control for elastic traffic," Proc. ITC Specialist Seminar, 1998. M. Nabe, M. Murata, and H. Miyahara, "Analysis and modeling of World Wide Web traffic for capacity dimensioning of Internet access lines," Perfbrmance Evaluation, Vbl. 34, No. 4, 1998. A. Riedl, T. Bauschert, M. Perske, and A. Probst, "Investigation of the M / G / R processor sharing model for dimensioning of IP access networks with elastic traffic," Proc. First Polish-German Teletraffic Symposium, 2000 . Fukushima, Nakamura, Nomoto, "Analysis of Multiple TCP Connection Resource Sharing Model Considering Slow Start," IEICE Technical Report, IN2002-29, pp. 43-48, 2002. Kawahara, Ishibashi, Asaka, "Bandwidth design management method using flow statistics," IEICE Technical Report, NS2003-87, pp. 1-4, 2003.

  The problem to be solved by the present invention is that, as described in the background art, in the conventional communication quality design management method, the communication bandwidth of a link in which flows having various access line bandwidths and round trip times are multiplexed. Design and communication quality management is difficult.

  The object of the present invention is to manage the bandwidth by associating the TCP quality of the flow of interest and the link usage rate, as well as to the case where flows having various access line bandwidths and round trip times are multiplexed. To provide a communication band design management method and a communication band design management apparatus capable of calculating a band required to satisfy each required quality, designing a communication band of the link, and managing the communication quality. .

  The above and other objects and novel features of the present invention will be apparent from the description of this specification and the accompanying drawings.

  The outline of the invention disclosed in the present application will be described as follows.

  (1) A communication bandwidth design and management method for designing and managing the communication quality of a user flow via a link in a communication network, where C is the communication bandwidth of the link, and r ′ is the transferable rate of the TCP flow of interest. The ratio R of the communication band C and the transferable speed r ′ R = C / r ′, the link usage rate ρ, and the TCP flow quality degradation frequency based on the quality of the TCP flow when the link usage rate is sufficiently small A communication bandwidth design management method for estimating the quality when the link usage rate becomes ρ using the quality degradation function D = D (R, ρ) representing the relationship of D and managing the communication quality of the link It is.

(2) In the means of (1), the maximum load assumed to be applied to the link is Lmax [bps], and the quality degradation degree function D is used as an allowable value ε for a predetermined quality degradation frequency ( 1 + ε) = D (C ^* / r ′, Lmax / C ^* ) is calculated so that the communication band C ^* is calculated and the communication band of the link is designed to be C ^*, and the TCP transfer rate of the link is r Maintain the communication quality of the user flow.

  (3) In the means of (1) or (2), the transferable speed r ′ of the TCP flow focused on the link is set to the access line bandwidth r of the flow or the maximum window size in the TCP communication of the flow. The smaller of the ratio Wmax / RTT between Wmax and the round trip time RTT and the value of the access line bandwidth r is the smaller.

を用いた下記数式２から算出する。

(4) In the means of (1) to (3), the quality deterioration degree function D = D (R, ρ) is expressed as C (R, ρ) expressed by the following formula 1.

It is calculated from the following formula 2 using

  (5) In the means of (1) to (4), instead of the quality degradation frequency D, the average file size of the flow of interest is s, and the TCP file transfer time when the link usage rate is sufficiently small is Calculated by s / r ′, and the average file transfer time T (ρ) when the link usage rate becomes ρ, T (ρ) = s / r ′ (R, ρ) is used to communicate the link. Manage quality.

(6) In the means of (2) to (4), instead of the quality degradation frequency D, the average file size of the flow of interest is s, and the TCP file transfer time when the link usage rate is sufficiently small is The maximum load that is calculated by s / r ′ and assumed to be applied to the link is Lmax [bps], and an allowable value T ^* for a predetermined average file transfer time is T ^* = s / r ′ × D (C ^{* / r ', Lmax / C} *) to calculate the C ^* satisfying the communication bandwidth of the link is designed such that C ^*.

(7) In the means of (2) to (6), when flows having various transferable speeds (classes) are multiplexed on the link, the transferable speeds of flows belonging to class i are set to r ′ ( i), the generation rate of class i flows is λ (i) [flows / s], the average file size of class i flows is s (i), and the allowable value ε (i ) Is set in advance and (1 + ε (i)) = D (C ^* (i) / r ′), Σλ (i) s (i) / C ^* (i) using the quality degradation function D ) Satisfying the communication bandwidth C ^* (i) for class i, and with C ^* = max {C ^* (i)}, the quality degradation degree D (i) of each class is D (i) ≦ (1 + ε (i)) is calculated as the communication bandwidth of the link necessary to satisfy (i)), and the communication bandwidth of the link is designed to be C ^* .

(8) In the means of (7), instead of the allowable value ε (i) for the quality degradation degree of the class i flow, an allowable value T ^* (i) for the average file transfer time of the class i flow is set in advance. Using the quality degradation degree function D, T ^* (i) = s (i) / r ′ (i) × D (C ^* (i) / r ′), Σλ (i) s (i ) / C ^* (i)) satisfying the bandwidth C ^* (i) for class i, and with C ^* = max {C ^* (i)}, the average file transfer time T (i) for each class is T (i) ≦ T ^* Calculated as the link communication band necessary to satisfy (i), and designed so that the communication band of the link is C ^* .

  (9) In the means of (7) or (8), the transferable rate r ′ (i) of the class i flow is set to the access line band r (i) of the class i flow or the class i flow. The smaller of the maximum window size Wmax (i) in TCP communication and the ratio Wmax (i) / RTT (i) of the round trip time RTT (i) and the access line bandwidth r (i).

(10) A communication band design management apparatus for designing and managing the communication quality of a user flow via a link in a communication network, wherein the communication band of the link is C, and the transferable speed of the TCP flow of interest is r ′ The ratio R of the communication band C and the transferable speed r ′ R = C / r ′, the link usage rate ρ, and the TCP flow quality degradation frequency based on the quality of the TCP flow when the link usage rate is sufficiently small It is assumed that the quality degradation degree function D = D (R, ρ) representing the relationship of D is used to estimate the quality degradation degree or average file transfer time of a flow belonging to the class of interest, and to be added to the link. Communication satisfying (1 + ε) = D (C ^* / r ′, Lmax / C ^* ) using the quality degradation function D as a maximum load Lmax [bps] and an allowable value ε for a predetermined quality degradation frequency communication bandwidth design management and means for calculating the band C ^* It is the location.

  The communication bandwidth design management method of the present invention, as in the means (1) to (9), determines the communication quality from the communication bandwidth C of the link, the transferable speed r ′ of the TCP flow of interest, and the link usage rate ρ. The degree of deterioration or the average file transfer time is calculated, the communication quality of the link is managed, and the communication band is designed. Even if a plurality of classes are mixed in one link, the communication band can be designed and managed so as to satisfy the communication quality required for each class.

  Further, when the communication band of the link is designed using the means of (1) to (9) and the communication quality is managed, for example, a communication band design management apparatus such as the means of (10) is targeted. Insert it into the link that becomes.

Hereinafter, the present invention will be described in detail together with embodiments (examples) with reference to the drawings.
Note that components having the same function are denoted by the same reference symbols throughout the drawings for describing the embodiments, and the repetitive description thereof is omitted.

  In the communication bandwidth design management method of the present invention, the degree of communication quality degradation or the average file transfer time is calculated from the communication bandwidth C of the link, the transferable speed r ′ of the TCP flow of interest, and the link usage rate ρ, and the link The communication band is designed and managed so as to satisfy the communication quality required for each class even if a plurality of classes are mixed in one link.

前記数式３において、D(R,ρ)はTCP品質劣化度であり、リンク使用率が十分小さいときの平均ファイル転送時間T_idealに対する、リンク使用率がρとなったときのファイル転送時間T(ρ)の劣化度合いを意味し、D(R,ρ)＝T(ρ)／T_idealで定義される。なお、T_idealは、平均ファイルサイズsを用いてT_ideal＝s／rで与えられる。また、前記数式３において、C(R,ρ)はアーランC式として知られているものであり、下記数式４で表される。

Among conventional communication bandwidth design management methods, for example, the method described in Non-Patent Document 7 assumes that the round trip time and access line bandwidth of each flow are uniform, and the influence of TCP slow start. The degree of TCP quality degradation when transferring a sufficiently large file is calculated using the parameter R given by R = (aggregated line bandwidth C) / (access line bandwidth r) by the following formula 3. .

In Equation 3, D (R, ρ) is a TCP quality degradation degree, and the file transfer time T (when the link usage rate becomes ρ with respect to the average file transfer time T _ideal when the link usage rate is sufficiently small. Denotes the degree of degradation of ρ) and is defined as D (R, ρ) = T (ρ) / T _ideal . Note that T _ideal is given by T _ideal = s / r using the average file size s. Further, in Equation 3, C (R, ρ) is known as an Erlang C equation and is represented by Equation 4 below.

  However, in general, since the access band of each user is various, the parameter R cannot be set by (aggregated line band C) / (access line band r).

なお、S(i)はクラスiの平均ファイルサイズとする。以下、前記近似式（数式５）の推定精度を評価した例を示す。 Therefore, in the present invention, when the flows having different access line bandwidths are multiplexed, the access line bandwidth of the class i flow is r (i), and the parameter R and the quality degradation degree D ( i) As an approximate expression of the average file transfer time T (i), the following expressions 5 are proposed.

S (i) is the average file size of class i. Hereinafter, an example in which the estimation accuracy of the approximate expression (Formula 5) is evaluated will be described.

  1 and 2 are schematic diagrams for explaining the principle of the communication band design management method of the present invention, and FIG. 1 is a graph showing an example of evaluation of estimation accuracy of an approximate expression used for communication band design and management. FIG. 2 is a diagram showing another evaluation example.

を満たす最小の正の整数Iに対して、クラスiフローの転送速度A(i,t)が下記数式７

で与えられるとしてシミュレーションを実施した。またこのとき、前記集約回線帯域C＝22.5Mbpsとし、４つのクラス（r(1)＝0.5Mbps，r(2)＝1.5Mbps，r(3)＝2.5Mbps，r(4)＝4.5Mbps）のフローが混在しているとする。また、ファイルサイズは、s(1)＝1MB，s(2)＝3MB，s(3)＝5MB，s(4)＝9MBとし、各クラスのフロー発生率を振らせてリンク使用率を変えたときの、各クラスの平均ファイル転送時間を評価した。なお、各クラスのフロー発生率の比は、（クラス１）：（クラス２）：（クラス３）：（クラス４）＝0.5：0.278：0.167：0.055となるように設定した。このような方法で評価を行った結果は、図１に示すようになり、高いアクセス回線帯域を持つフローの転送時間を精度良く推定できていることが分かる。また、低いアクセス回線帯域フローについては、若干大きめに近似しているが安全側となっている。 In order to evaluate the estimation accuracy of the approximate expression represented by Equation 5, as an example of the evaluation, the transfer rate of the TCP flow in communication at a certain time is equal to the aggregated line bandwidth C by the number of simultaneously connected flows at that time. A simulation was performed so that the divided value and the access line bandwidth of the flow of interest were smaller. Specifically, the number of flows of class i at time t is n (i, t), and the sum is n (t) = Σn (i, t), r (1) <r (2) <... <r ( k) (k is the number of classes)

The transfer rate A (i, t) of the class i flow for the smallest positive integer I satisfying

The simulation was carried out as given in At this time, the aggregated line bandwidth C = 22.5 Mbps, and four classes (r (1) = 0.5 Mbps, r (2) = 1.5 Mbps, r (3) = 2.5 Mbps, r (4) = 4.5 Mbps) Are mixed. The file size is s (1) = 1MB, s (2) = 3MB, s (3) = 5MB, s (4) = 9MB, and the link usage rate is changed by changing the flow rate of each class. The average file transfer time for each class was evaluated. In addition, the ratio of the flow rate of each class was set to be (class 1) :( class 2) :( class 3) :( class 4) = 0.5: 0.278: 0.167: 0.055. The result of evaluation by such a method is as shown in FIG. 1, and it can be seen that the transfer time of a flow having a high access line bandwidth can be estimated with high accuracy. Also, the low access line bandwidth flow is on the safe side although it is a little larger.

In the approximate expression expressed by Equation 5, the influence of TCP slow start can be ignored, but it cannot always be ignored when the file size to be transferred is small. Therefore, instead of the access line band r (i), assuming that an actual transferable speed r ′ (i) considering TCP slow start is given, an approximation represented by the following formula 8 instead of the formula 5 is given. An expression may be used.

  In order to evaluate the estimation accuracy of the approximate expression represented by Equation 8, a simulation simulating TCP operation was performed. At this time, it is assumed that the aggregated line bandwidth C = 10 Mbps, the flow with an access line bandwidth of 2 Mbps and the flow with 600 kbps are mixed, and the average file size is 10 KB. We also evaluated the average file transfer time when the ratio of the flow rate of 2Mbps and the flow rate of 600kbps was 1: 2, and the link usage rate was changed by changing the flow rate. For the transferable speed r ′ of the flow in the approximate expression, the average file transfer time T ′ when the link usage rate is sufficiently small is measured for each class, and the value obtained by dividing the file size 10 KB by T ′ is used as r ′ ( R 'for the 2Mbps flow was 1.5288Mbps and r' for the 600kbps flow was 575.51kbps). The evaluation result at this time is, for example, as shown in FIG. 2, and it is understood that the TCP quality of each class can be accurately estimated by the proposed approximate expression.

  As described above, even if multiple classes are mixed, the bandwidth ratio R (i) = C / r (i) is obtained for each class using the access line bandwidth (or transferable speed) of the class of interest. Can be used instead of R in the uniform class estimation formula to estimate the TCP quality for each class.

  Therefore, in the communication band design management method of the present invention, the link communication band is C, the transferable speed of the focused TCP flow is r ′, the ratio thereof is R = C / r ′, the link usage rate ρ, the link usage When the link usage rate becomes ρ using the quality degradation function D = D (R, ρ) that represents the relationship of the quality degradation frequency D of the TCP flow based on the TCP flow quality when the rate is sufficiently small Estimate the quality of communication and manage the communication quality.

  At this time, the transferable speed r ′ of the TCP flow is set to, for example, the access line bandwidth, or the transferable speed r ′ is set by measuring the file transfer time when the link usage rate is sufficiently small. As another setting method, the transferable speed r ′ of the TCP flow of interest is the ratio of the access line bandwidth r of the flow or the maximum window size Wmax in the TCP communication of the flow and the round trip time RTT. The smaller of Wmax / RTT and the value of the access line band r is assumed. Here, since Wmax / RTT is the maximum transferable speed in the TCP window control, the smaller transfer line speed r is the transferable speed.

Further, in the communication band design management method of the present invention, the maximum load assumed to be applied to the link is Lmax [bps], and the allowable value ε for a predetermined quality deterioration frequency is used, and the quality deterioration frequency D is used. (1 + ε) = D ( C * / r ', Lmax / C *) to calculate the C ^* satisfying the link bandwidth is designed to be C ^*.

At this time, the allowable value ε is a parameter for determining how much quality (bandwidth) the quality of communication service is to be provided. For example, if ε is increased, the necessary bandwidth C ^* can be reduced. Therefore, this leads to cost reduction, but a large degree of quality degradation is allowed, and if customer satisfaction cannot be obtained, there is a risk of reducing the number of customers. Therefore, it is a parameter determined by a trade-off between service level and cost.

3 and 4 are schematic diagrams for explaining a communication band design management method according to an embodiment of the present invention. FIG. 3 is a diagram showing an example of an IP network to which the present invention is applied. FIG. It is a figure which shows schematic structure of a band design management apparatus.
In FIG. 3, 1A and 1B are nodes, 2 is a link, 3 is a bandwidth design management device, 4A and 4B are terminals, 5A is a class 1 user terminal, and 5B is a class 2 user terminal. In FIG. 4, 301 is a packet header analysis unit, 302 is a flow quality measurement unit, 303 is a link usage rate calculation unit, 304 is a flow quality / use rate management unit, and 305 is a bandwidth design unit.

  A communication network to which the communication quality management method of the present invention is applied is a network composed of a plurality of nodes and links, such as an IP network. For example, as shown in FIG. 3, between nodes 1A and 1B An apparatus (communication band design management apparatus) 3 for designing and managing the communication band of the link 2 is inserted into the link 2 connecting the two. At this time, the link 2 into which the communication band design management device 3 is inserted is, for example, from the terminals 4A and 4B on the network, the user terminal 5A having an access line of 2 Mbps (class 1), and a user having 600 kbps (class 2) Assume that this is an aggregated line through which packets (user flows) transmitted to the terminal 5B pass. At this time, in the example shown in FIG. 4, each of the terminals 4A, 4B, 5A, 5B is one, but a plurality of the terminals 4A, 4B, 5A, 5B may be provided. Further, a network may be used in which packets are transmitted via a plurality of nodes and links between the terminals 4A and 4B and the node 1A on the network or between the node 1B and the user terminals 5A and 5B.

  At this time, for example, as shown in FIG. 4, the communication band management device 3 measures the packet header analysis means 301 for analyzing a packet arriving from the preceding node 1A, and the communication quality of the flow to be managed. Flow quality measuring means 302, link usage rate calculating means 303 for calculating the usage rate of the link to be managed, flow quality / usage rate management means 304 for managing the communication quality and link usage rate of the flow, and communication of the flow Band design means 305 for designing a band necessary for maintaining the quality is provided.

  Hereinafter, an example of a communication band design method and a communication quality management method for the link 2 using the communication band design management device 3 will be described.

  When a packet arrives at the communication bandwidth design management device 3 from the preceding node 1A, the communication bandwidth design management device 3 adds, for example, the source IP to the packet header analysis unit 301 in addition to the packet size and TCP flag information. Information such as address, destination IP address, source port number, destination port number, and protocol number is read and transmitted to the subsequent node 1B. At this time, the packet header analysis unit 301 prepares in advance a table for associating information such as an address made up of the set of the five information with a service class, determines the class L of the arrived packet, It is determined whether it is a flow to be performed. At this time, the correspondence table may be held by the packet header analysis unit 301, or may be held separately from the packet header analysis unit 301. Further, instead of determining the class by referring to the table, a flag indicating the class is set in the header for each packet at the entrance of the network, and the packet header analysis unit 301 determines the class from the flag in the header. It may be determined.

  Further, when associating the arrived packet with a class, it may be, for example, a set of two pieces of information of a source IP address and a destination IP address instead of the five pieces of information, and depends on how the class is defined. Thus, the class discrimination method is determined.

  At this time, the number of flows to be managed may be one or plural, but here, for the sake of simplicity, it is assumed that there is one flow to be managed.

  When the packet header analysis unit 301 determines that the arrived packet belongs to the management target flow, the TCP flag information and the packet size at that time are notified to the flow quality measurement unit 302.

When the flow quality measuring unit 302 detects a packet in which the SYN flag bit indicating the start of the TCP flow is set in the notified TCP flag, it determines that a new flow has occurred, and the time at that time is a variable T _start. Remember it. On the other hand, if a packet with the FIN flag bit indicating the end of the TCP flow is detected, the flow is completed, and a value obtained by subtracting T _start from the current time is calculated and measured as the file transfer time Tf (n). To do. Here, n means the nth detected flow.

  In addition, the flow quality measuring means 302 measures the file transfer time Tf (n) and calculates the sum of the number of bytes of packets transferred between the detection of the SYN flag bit and the detection of the FIN flag bit. Measured as file size Sf (n). This measurement is performed only during a limited time period when the link usage rate is small. After the measurement, the average avgTf = ΣTf (n) / N and avgSf = ΣSf (n) / N (N is the number of measurements) are calculated, and the result is notified to the flow quality / usage management means 304. In this example, the packet flowing through the link is captured and the flow communication quality is measured. Instead, a test packet for measuring the flow communication quality is sent from a certain packet generation device to the packet reception device. You may send and measure quality.

  The packet header analysis unit 301 notifies the link usage rate calculation unit 303 of all read packet sizes regardless of the flow ID. The link usage rate calculation means 303 divides the sum of the sizes of packets passed during a predetermined time t0 by t0, and further divides it by the link communication band C to obtain the link usage rate ρ (j) for each t0. This is calculated and notified to the flow quality / usage rate management means 304 every t0 hours.

  The flow quality / usage rate management unit 304 uses the avgTf and avgSf received from the flow quality measurement unit 302 to calculate the transferable speed r ′ of the flow by r ′ = avgSf / avgTf. Then, a bandwidth ratio R = C / r ′ is calculated from the transferable speed r ′ and the communication bandwidth C of the link, and the j-th measured link usage rate ρ (j ) And R are substituted into the quality degradation function D (R, ρ (j)) expressed by the above Equation 3, and the quality degradation level D (j) at that time is calculated. At this time, if the quality degradation degree D (j) exceeds a predetermined allowable quality value 1 + ε, it is determined that the quality is degraded. If it is determined that the quality is deteriorated, the calculated transferable speed r ′ is notified to the bandwidth design unit 305.

In the bandwidth design unit 305, the value of the transferable speed r ′ notified from the flow quality / usage rate management unit 304, the maximum load Lmax [bps] assumed to be applied to the link, and predetermined quality degradation A communication band C ^* satisfying (1 + ε) = D (C ^* / r ′, Lmax / C ^* ) is calculated using the allowable value ε for the frequency and the quality degradation degree function D expressed by Equation 3 above. Then, the calculated communication band C ^* is set as a new band of the band in which the quality is deteriorated.

  As described above, according to the communication bandwidth design management method of the present embodiment, the communication bandwidth C of the link, the transferable speed r ′ of the TCP flow of interest, the link usage rate ρ, the degree of communication quality degradation, or the average By calculating the file transfer time, managing the communication quality of the link, and designing the communication bandwidth, the communication bandwidth satisfies the communication quality required for each class even if multiple classes are mixed in one link. Can be designed and managed.

In the communication band design management method of the present embodiment, the flow quality / usage management means 304 uses the quality degradation level D (j) as a quality measure, but instead of the quality degradation level D (j), As shown in Equation 5 or Equation 8, the average file transfer time may be used as a quality measure. In this case, the average file transfer time T (j) of the flow of interest when the link usage rate is ρ (j) is calculated by T (j) = avgTf × D (R, ρ (j)). Is compared with a predetermined allowable quality value T ^*, and if T (j)> T ^*, it is determined that the quality is deteriorated.

At this time, the bandwidth design unit 305 uses the maximum load Lmax [bps] assumed to be applied to the link and uses a predetermined allowable value T ^* for the average file transfer time, T ^* = avgTf A communication band C ^* satisfying × D (C ^* / r ′, Lmax / C ^* ) is calculated.

  Further, the flow transferable speed r ′ may be set as the flow transferable speed r ′ instead of being set through actual measurement as in this embodiment. In addition, for example, the ratio Wmax / RTT between the maximum window size Wmax and the round trip time RTT in TCP communication of the flow, and the smaller value of the access line r is set as the flow transferable speed r ′. Also good.

Further, in the bandwidth design unit 305, instead of using the assumed maximum load Lmax [bps] as in this embodiment, the transferable speed of the flow belonging to the class i multiplexed on the link is set to r ′ (i ), The generation rate of class i flows is λ (i) [flows / s], the average file size of class i flows is s (i), and the allowable value ε (i) for the quality degradation level of class i flows is set in advance. In addition, (1 + ε (i)) = D (C ^* (i) / r ′ (i), Σλ (i) s (i) / C using the quality deterioration degree function D expressed by the above-mentioned equation 3. ^* (i)) is calculated band C ^* a (i) for the class i satisfying, C ^* = max {C ^* (i)} with each class quality deterioration degree D (i) is D (i) ≦ It may be calculated as a communication band necessary to be (1 + ε (i)).

Further, the bandwidth design unit 305 may use the average file transfer time as a quality measure instead of using the quality degradation degree D (i) as a quality measure. In this case, an allowable value T ^* (i) for the average file transfer time of the class i flow is set in advance, and T ^* (i) = s (i) using the quality deterioration degree function D expressed by Equation 3 above. ) / R ′ (i) × D (C ^* (i) / r ′ (i), Σλ (i) s (i) / C ^* (i)) satisfying band C ^* (i) for class i And C ^* = max {C ^* (i)} is calculated as a communication band necessary for the average file transfer time T (i) of each class to satisfy T (i) ≦ T ^* (i).

  The present invention has been specifically described above based on the above-described embodiments. However, the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the scope of the present invention. is there.

It is a schematic diagram for demonstrating the principle of the communication band design management method of this invention, and is a graph which shows an example of evaluation of the estimation precision of the approximate expression used for the design and management of a communication band. It is a schematic diagram for demonstrating the principle of the communication band design management method of this invention, and is a figure which shows the other evaluation example. It is a schematic diagram for demonstrating the communication band design management method of one Example by this invention, and is a figure which shows an example of the IP network to which this invention is applied. It is a schematic diagram for demonstrating the communication band design management method of one Example by this invention, and is a figure which shows schematic structure of a communication band design management apparatus.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1A, 1B ... Node 2 ... Link 3 ... Communication band design management apparatus 301 ... Packet header analysis means 302 ... Flow quality measurement means 303 ... Link usage rate calculation means 304 ... Flow quality / use rate management means 305 ... Band design means 4A, 4B: Terminal on network 5A: Class 1 user terminal 5B: Class 2 user terminal

Claims (10)

A communication band design management method for designing and managing communication quality of user flows via a link in a communication network,
The communication bandwidth of the link is C, the transferable speed of the TCP flow of interest is r ′, the ratio R = C / r ′ of the communication band C and the transferable speed r ′, the link usage rate ρ, and the link usage rate When the link usage rate becomes ρ using the quality degradation function D = D (R, ρ) that represents the relationship of the TCP flow quality degradation frequency D based on the quality of the TCP flow when is sufficiently small A communication band design management method characterized by estimating the quality of the link and managing the communication quality of the link. The maximum load assumed to be applied to the link is Lmax [bps], and an allowable value ε for a predetermined quality deterioration frequency is set to (1 + ε) = D (C ^* / r ′, Lmax / C ^*) to calculate the communication band C ^* satisfying the communication bandwidth of the link is designed to be C ^*, TCP transferable speed of the link to maintain the communication quality of user flow to be r ' The communication band design management method according to claim 1.   The transferable speed r ′ of the TCP flow of interest in the link is determined by determining the access line bandwidth r of the flow or the ratio Wmax / RTT of the maximum window size Wmax and the round trip time RTT in the TCP communication of the flow and the access line. The communication band design management method according to claim 1 or 2, wherein the band r is set to a smaller value. The quality degradation function D = D (R, ρ) is expressed by the following formula 1 as C (R, ρ)

The communication bandwidth design management method according to claim 1, wherein the communication bandwidth design management method is calculated from the following formula 2 using:

  Instead of the quality degradation frequency D, the average file size of the flow of interest is s, the TCP file transfer time when the link usage rate is sufficiently small is calculated by s / r ′, and the link usage rate is ρ The communication quality of the link is managed by using T (ρ) = s / r ′ × D (R, ρ) as an average file transfer time T (ρ) at that time. The bandwidth design management method according to claim 4. Instead of the quality degradation frequency D, it is assumed that the average file size of the flow of interest is s, the TCP file transfer time when the link usage rate is sufficiently small is calculated by s / r ′, and added to the link C ^* satisfying T ^* = s / r ′ × D (C ^* / r ′, Lmax / C ^* ), where Lmax [bps] is the maximum load and the allowable value T ^* for a predetermined average file transfer time. 5. The communication band design management method according to any one of claims 2 to 4, wherein the communication band is calculated and designed so that a communication band of the link is C ^* . When flows with various transferable speeds (classes) are multiplexed on the link,
Let r '(i) be the transferable speed of flows belonging to class i, λ (i) [flows / s] the generation rate of flows of class i, s (i) the average file size of flows of class i, and class An allowable value ε (i) for the quality deterioration degree of the flow of i is set in advance, and (1 + ε (i)) = D (C ^* (i) / r ′ (i) using the quality deterioration degree function D , Shigumaramuda the (i) s (i) / C * (i)) satisfying the communication band C ^* (i), is calculated for the class ^{i, C * = max {C} * (i)} to have in each class The quality degradation degree D (i) is calculated as the communication bandwidth of the link necessary for D (i) ≦ (1 + ε (i)), and the communication bandwidth of the link is designed to be C ^*. The communication band design management method according to any one of claims 2 to 6, wherein the communication band design management method is characterized. Instead of the allowable value ε (i) for the quality degradation level of the class i flow, an allowable value T ^* (i) for the average file transfer time of the class i flow is set in advance, and the quality degradation level function D T ^* (i) = s (i) / r ′ (i) × D (C ^* (i) / r ′ (i), Σλ (i) s (i) / C ^* (i)) The bandwidth C ^* (i) to be satisfied is calculated for class i, and the average file transfer time T (i) for each class is T (i) ≦ T ^* (i) with C ^* = max {C ^* (i)} The communication band design management method according to claim 7, wherein the link communication band is calculated as the link communication band necessary to satisfy the condition, and the link communication band is designed to be C ^* .   The transferable speed r ′ (i) of the class i flow is the access line bandwidth r (i) of the class i flow, or the maximum window size Wmax (i) in the TCP communication of the class i flow, and 9. The communication according to claim 7, wherein the ratio of the round trip time RTT (i) Wmax (i) / RTT (i) and the value of the access line bandwidth r (i) is the smaller one. Bandwidth design management method. A communication bandwidth design management device for designing and managing communication quality of user flows via a certain link in a communication network,
The communication bandwidth of the link is C, the transferable speed of the TCP flow of interest is r ′, the ratio R = C / r ′ of the communication band C and the transferable speed r ′, the link usage rate ρ, and the link usage rate The quality degradation degree of a flow belonging to the class of interest using a quality degradation degree function D = D (R, ρ) representing the relationship of the TCP flow quality degradation degree D based on the quality of the TCP flow when is sufficiently small Alternatively, the means for estimating the average file transfer time, and the maximum load assumed to be applied to the link is Lmax [bps], and the quality deterioration degree function D is used as an allowable value ε for a predetermined quality deterioration degree ( 1 + ε) = D (C ^* / r ′, Lmax / C ^* ). A communication band design management apparatus comprising means for calculating a communication band C ^* that satisfies L ^* / C ^* .

Images