JP2011211358A - Apparatus for estimating quality degradation spot in network - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a quality degradation spot estimation apparatus for a network which can accurately estimate a quality degradation spot even when accurate communication route information of each path can not be acquired from the network.SOLUTION: A measurement part 101 acquires measured results of a packet loss rate X1 and packet delay variation X2 in each path by instructing respective end nodes. A path classification part 102 classifies respective paths into a plurality of groups including a defective path group and an excellent path group in respect to quality degradation of respective paths. A network division part 103 divides a network to be evaluated into a plurality of candidate spots each of which is a minimum unit capable of estimating quality degradation. An identification part 104 identifies candidate spots on which only excellent paths pass. A totalization part 105 totalizes the number of passages of defective paths for remaining candidates spots excluding the candidate spots on which only the excellent paths pass. An estimation part 106 estimates defective spots out of the candidate spots based on the number of passages of the defective paths.

Description

  The present invention relates to an apparatus for estimating a network quality degradation location, and in particular, a quality degradation location estimation device suitable for network quality degradation location estimation in which a plurality of paths are established on the same path for the purpose of load distribution and multiplexing. About.

  As a method of identifying quality degradation locations in the network in real time, end-to-end packet loss rate and packet delay are periodically measured on multiple paths, and quality degradation is based on the correlation between measured data and path information. Non-Patent Documents 1 and 2 disclose methods for estimating the location.

  Non-Patent Document 1 is an estimation method based on the packet loss rate, and compares the packet loss rate of each path with a preset threshold value for each measurement period to classify the path status into three (good / bad / medium). The packet loss occurrence location is estimated by mapping the classification result to the network topology information.

  Non-Patent Document 2 is an estimation method based on packet delay. Each path is classified into a plurality of clusters (sets) focusing on the similarity of time-series data of packet delay, and the classification result is mapped to network topology information By doing so, the location where the packet delay occurs is estimated.

  FIG. 11 is a diagram schematically illustrating a method for estimating a quality degradation location according to Non-Patent Document 2. Here, regarding a path 1 in which a transmission / reception node is a pair of nodes A and B, a communication path is Ra → Rf →. Measurement data is obtained in which Re → Rd, the quality classification based on the packet loss rate is “bad”, and the quality classification based on the packet delay is “cluster 1”.

  As for the other paths 2-6, if the measurement data is as shown in the figure, Rf → Re is identified as a link common to the paths whose classification based on the packet loss rate is “bad”, but in the opposite direction. The two paths 5 and 6 including the link Re → Rf are classified as “good” based on the packet loss rate, and the classification based on the packet delay is also “cluster 0”, so the link quality of Rf → Re is degraded. It is estimated that

A. Tachibana, S. Ano, T. Hasegawa, M. Tsuru, Y. Oie, "Locating Congested Segments over the Internet Based on Multiple End-to-End Path Measurements," IEICE Transactions 89-B (4), pp. 1099-1109 (2006). A. Tachibana, S. Ano, T. Hasegawa, M. Tsuru, Y. Oie, "Locating congested segments over the Internet by clustering the delay performance of multiple path," Computer Communications 32, pp. 1642-1654 (2009).

  Each of the above prior arts is based on the premise that communication path information of each path can be collected from the network. Therefore, there is a problem of erroneous estimation if the communication path information cannot be collected. Especially, in recent years, it is difficult to obtain accurate communication path information for each path in a network where multiple communication paths (candidates) are set on the same path to achieve load balancing and improve reliability. There are many.

  FIG. 12 is a diagram illustrating an example of a network that employs load balancing. In the illustrated example, the node Ra autonomously links to either one (or both) of the two nodes Rb and Rf. However, since it is difficult to obtain information on what kind of route is established between nodes A and B from the network, even if quality degradation occurs in the paths of nodes A and B The specific quality location cannot be estimated.

  An object of the present invention is to solve the above-described problems of the prior art and provide a network quality degradation location estimation device that can accurately estimate a quality degradation location even when accurate communication path information of each path cannot be obtained from the network. There is.

  In order to achieve the above object, the present invention provides a measurement for measuring an index value of quality degradation for each path in a network quality degradation location estimation apparatus in which a path established between end nodes includes at least one route candidate. Means were provided. As this index value, for example, the packet loss rate X1 and the packet delay variation X2 can be used.

  The present invention is also capable of estimating the quality degradation of the path classification means for classifying each path into a plurality of groups including a group of bad paths and a group of good paths with respect to quality degradation based on the index value. Network dividing means for dividing into a plurality of candidate locations, which is the minimum unit, is provided. This network dividing means divides the network into, for example, a plurality of sections ki having no branching and merging, and a network device nj located at a branching / merging point of each section ki, and each section ki and network device nj are set as candidate locations.

  The present invention further adds an identification means for identifying candidate locations through which only the good path passes, and counts the number of times the defective path passes through the set of remaining candidate locations excluding candidate locations through which only the good path passes. Aggregating means and estimating means for estimating a defective portion based on the number of times the defective path passes are provided from the candidate portions.

According to the present invention, the following effects are achieved.
(1) Even in a network in which a plurality of communication path candidates exist in each path, the degradation location can be estimated when the quality of each path is degraded.
(2) When estimating a deteriorated part, not only a link but also a network device can be a candidate, so that the deteriorated part can be accurately estimated for each link and network device.

It is the figure which showed an example of the topology information of the network which this invention evaluates. It is the figure which showed the structure of each path | pass established between nodes. It is the flowchart which showed the estimation method of the quality degradation location in this invention. It is the figure which showed the method of dividing | segmenting a network into a some candidate location regarding the quality degradation location. 5 is a flowchart showing a method for estimating a defective portion based on a packet loss rate X1. It is the figure which showed an example of the method of excluding a part of candidate location from the set of candidate locations. It is the figure which showed the example of totalization of a bad path | pass. It is the figure which showed the example of estimation of the quality degradation location. 5 is a flowchart showing a method for estimating a deterioration location based on packet delay variation X2. It is a functional block diagram of the quality degradation location estimation apparatus which concerns on this invention. It is the figure which expressed typically the estimation method of the quality degradation location by a nonpatent literature 2. FIG. It is the figure which showed an example of the network which employ | adopts load balancing.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a diagram showing an example of network topology information to be evaluated by the present invention. A quality degradation location device 1 that estimates a quality degradation location is connected via a network device (in this case, Rc) such as a router. Connected to the network. FIG. 2 is a diagram showing a path configuration established between the nodes. In the present embodiment, one path is a plurality of path candidates L1, L2 like the path 1 from the node A to the node B. May be included.

FIG. 3 is a flowchart showing a method for estimating a quality deterioration point according to an embodiment of the present invention, and mainly shows operations of the quality deterioration point device 1 and the end nodes A, B, C. In the present embodiment, the quality degradation point is estimated by the following five steps, and the estimation process is repeatedly executed at a predetermined cycle.

  (1) Step S1: The end-to-end packet loss rate X1 and the packet delay variation X2 are actively measured simultaneously on all paths.

  (2) Based on the measurement result, it is determined whether or not there is a path with degraded quality. If there is a degraded path, the process proceeds to step S3 and subsequent steps to estimate the degraded location.

  (3) Step S3: The measurement results of the packet loss rate X1 and the packet delay variation X2 are analyzed, and the quality information of each path is calculated (clustering).

  (4) Step S4: The network is divided into a plurality of candidate locations (interval ki and network device nj described later) with respect to quality degradation locations.

(5) Step S5: A quality degradation point is estimated by mapping the quality information of each path to the topology information.

  In step S1, for each path, test packets are continuously transmitted and received at independent random time intervals, and the end-to-end packet loss rate X1 and the packet delay variation X2 are simultaneously used as indices representing the communication quality of each path. It is measured. Such measurement is performed by each end node performing measurement in response to a measurement request from the quality degradation location device 1 and providing a measurement result to the quality degradation location device 1. The packet delay variation X2 is defined by a delay variation excluding a fixed processing time such as a transmission delay or a router from a time when a packet arrives from the transmission side to the reception side (packet delay). , Packet delay-approximated by the minimum packet delay within the period.

  In step S2, it is determined in the quality degradation location device 1 whether or not a degraded path is detected in which at least one of the packet loss rate X1 and the packet delay variation X2 exceeds preset thresholds X1ref and Xref2. If a path with degraded quality is detected, the process proceeds to step 3 and subsequent steps to estimate the quality degraded location. If a path with degraded quality is not detected, the quality degraded location is not estimated in the corresponding period.

  If at least one degraded path with a packet loss rate X1 exceeding the threshold value X1ref is detected in step 2, in step S3, the packet loss rate X1 of all paths is preset in the quality degradation location device 1. Compared with two threshold values X1refl and X1refh, each path is classified into one of three states (good / medium / bad).

  In the present embodiment, the threshold values X1ref and X1refl are the same value, and X1refl <X1refh. Then, each path of X1 ≦ X1refl is classified into a good path group, each path of X1refl <X1 ≦ X1refh is classified into a medium path group, and each path of X1refh <X1 is classified into a bad path group. As described above, in the present embodiment, each path is classified into three states in order to avoid erroneous determination, and information on paths classified into the medium path group is not used for subsequent quality degradation point estimation.

  On the other hand, if at least one degraded path in which the packet delay variation X2 exceeds a preset threshold value X2ref (for example, 10 ms) is detected in step 2, in step S3, the packet delay variation X2 Based on the sequence similarity, each path is classified into a plurality of clusters.

  Thus, in this embodiment, a set of good / bad paths (Sgood, Sbad) is obtained as a path classification result based on the packet loss rate X1, and the same quality degradation is obtained as a path classification result based on the packet delay variation X2. A set of paths (clusters) C1, C2,..., Cn and a path set C0 that has not experienced quality degradation are obtained. Here, in the classification based on the packet loss rate X1, each path is classified into one of Sgood and (Smedium,) Sbad, but in the classification based on the packet delay variation X2, the same path is classified into a plurality of clusters. Sometimes it is done.

  In step 4, in the quality degradation location device 1, as shown in an example in FIG. 4, the network to be evaluated is the smallest unit that can estimate the quality degradation of the network on the basis of the branch / junction point of each path. Divided into a plurality of candidate locations. In the present embodiment, the network is divided into a plurality of sections ki (consisting of a network device and a link connecting the network devices), and a network such as a router located at a branch or junction of each section ki. The device is divided into devices nj, and each section ki and each network device nj are set as candidates for quality degradation. In the present embodiment, even if the constituent elements of the section ki are the same, they are treated as different sections if the packet passing directions are different.

  Returning to FIG. 3, in step S <b> 5, in the quality degradation location device 1, for the quality degradation candidate locations ki and nj, estimation of the failure location based on the packet loss rate X <b> 1 and estimation of the failure location based on the packet delay variation X <b> 2. Is done. Hereinafter, each of the method for estimating the defective portion based on the packet loss rate X1 and the method for estimating the defective portion based on the packet delay variation X2 will be described in detail.

[Estimation based on packet loss rate]
FIG. 5 is a flowchart showing a method for estimating a defective location based on the packet loss rate X1, and mainly shows the operation of the quality degradation location device 1.

  In step S21, one of the good paths is selected as the current focused good path. In step S22, with respect to the current good path of interest, a portion shared by all of the candidate routes is excluded from the set of quality deterioration candidate portions. In the example of FIG. 6, since path 1 includes two route candidates L1 and L2, each route candidate L1 and L2 shares a section of transmission node A → network equipment Ra and a section of network equipment Rd → receiving node B. The two sections k1 and k10 are excluded from the set of candidate locations. However, even in the sections k1 and k10, the reverse routes (network device Ra → transmission node A and reception node B → network device Rd) are handled differently and are not excluded from the set of candidate locations.

  In step S23, it is determined whether or not the above exclusion process has been completed for all good paths. If it is not completed, the process returns to step S21, and candidate portions are narrowed down by repeating the above processes for all good paths while switching the good path of interest.

  In step S24, the number of bad paths that may pass or pass through is counted for all the remaining locations (section ki and network device nj) as candidate locations for quality degradation. FIG. 7 is a diagram illustrating an example of totaling bad paths that pass through. If path 1 with end nodes A and B as transmission / reception nodes is a bad path, two route candidates L1 and L2 of the path pass. The count values of all the sections ki and the network device nj are incremented (+1). In step S25, it is determined whether or not the above totaling has been completed for all bad paths. If not completed, the process returns to step S24, and the counting is repeated while switching the bad path.

  In step S26, the ratio Tloss (= Nbpath / | Sbad |) of the number of bad paths Nbpath that may or may pass through each candidate location and the total number of bad paths (= | Sbad |) is calculated. The largest candidate location is estimated as a quality degradation location. FIG. 8 is a diagram showing an example of estimation of a quality degradation point. Here, since the count value of the section k6 from the network device Rf to the same Re is “4”, the maximum value is shown. Estimated as a point.

[Estimation method by packet delay variation]
FIG. 9 is a flowchart showing a method for estimating a degradation location based on the packet delay variation X2, and mainly shows the operation of the quality degradation location device 1. Here, the deteriorated part is estimated for each cluster (C1, C2... CN) obtained as a path classification result.

  In step S31, one of the clusters Ci (C0, C1,... Cn) is selected as the current attention cluster Ct. In step S32, for all the paths Pj that are not included in the current cluster of interest Ct, a shared location between route candidates of each path is identified and excluded from the set of candidate locations. In step S33, the number of times that the path included in the current cluster of interest Ct passes through the relevant part for all the remaining parts (section ki and network device nj included in the relevant part ki) as candidate places for quality degradation. Are counted.

  In step S34, a ratio Tdelay_i (= Ncpath / | Ct |) between the number of paths Ncpath passing through each candidate location and the total number of paths included in the current cluster of interest Ct (= | Ct |) is calculated, and Tdelay_ A candidate location where i is maximized is estimated as a quality degradation location. Note that the procedure of steps S31 to S34 is the same as that in the estimation based on the packet loss rate when the set of paths included in the current cluster of interest Ci is Sbad and the set of other paths is Sgood.

  In step S35, it is determined whether or not the above estimation procedure has been completed for all the clusters Ci. If not completed, the process returns to step S31, and the above estimation procedure is repeated for other clusters Ck (k = 1, 2... N, i ≠ k) while switching the class of interest Ct, and quality degradation is performed for each cluster Ci. The location is estimated.

  FIG. 10 is a functional block diagram of the quality degradation location estimation apparatus 1. The measurement unit 101 instructs each end node to measure an index value of quality degradation for each path, and the measurement result is obtained from each end node. get. In the present embodiment, measurement of the packet loss rate X1 and the packet delay variation X2 is instructed, and the measurement results are acquired.

  Based on the measurement result of the index value, the path classifying unit 102 classifies each path into a plurality of groups including a defective path group and a superior path group with respect to quality degradation. In this embodiment, for the packet loss rate X1, each path is classified into a bad path group, a good path group, and an intermediate path group based on the packet loss rate. Not used for estimation. As for the packet delay variation X2, each path is classified into a plurality of clusters based on the packet delay variation.

  The network dividing unit 103 divides the network to be evaluated into a plurality of candidate locations, which are minimum units that can estimate the quality degradation. In the present embodiment, the network is divided into a plurality of sections ki without branching and joining, and network equipment nj located at the branching and joining points of each section ki, and each section ki and network equipment nj are set as candidate locations.

  The identification unit 104 identifies candidate locations through which only the excellent path passes. The totaling unit 105 totalizes the number of times the defective path passes for a set of remaining candidate locations excluding candidate locations through which only the excellent path passes. The estimation unit 106 estimates a defective location from the candidate locations based on the number of times the defective path passes. In the present embodiment, a candidate location with the largest number of passing defective paths is estimated as a defective location.

  DESCRIPTION OF SYMBOLS 1 ... Quality degradation location estimation apparatus, 101 ... Measuring part, 102 ... Path classification | category part, 103 ... Network division | segmentation part, 104 ... Identification part, 105 ... Aggregation part, 106 ... Estimation part

Claims (6)

In the quality degradation location estimating apparatus for a network in which a path established between end nodes includes at least one route candidate,
A measuring means for measuring an index value of quality degradation for each pass;
Path classification means for classifying each path into a plurality of groups including a group of defective paths and a group of good paths based on the index value;
A network dividing means for dividing the network into a plurality of candidate locations, which is the smallest unit capable of estimating the quality degradation;
Identifying means for identifying candidate locations through which only the excellent path passes;
A counting means for counting the number of times that the defective path passes for the remaining candidate places excluding the candidate places where only the excellent path passes;
An apparatus for estimating a quality degradation location of a network, comprising: estimation means for estimating a defective location based on the number of times the defective path passes from among the candidate locations.   The network quality degradation location estimation apparatus according to claim 1, wherein the quality degradation index value is a packet loss rate.   The network quality degradation location estimation apparatus according to claim 1, wherein the quality degradation index value is a packet delay variation.   The classifying means classifies each path into a group of bad paths, a group of good paths, and a group of intermediate paths based on the packet loss rate, and the intermediate paths are not used for estimation of quality degradation points. The network quality degradation location estimation apparatus according to claim 2. The classifying unit classifies each path into a plurality of clusters based on packet delay variation,
The identifying means identifies candidate locations shared by route candidates for each path not included in the cluster for each cluster,
4. The network quality degradation location estimation apparatus according to claim 3, wherein the aggregation means counts the number of times a path included in the cluster passes through each candidate location for each cluster.   The network dividing unit divides the network into a plurality of sections without branching and joining, and a plurality of network devices located at the branching and joining points of each section, and each section and network device are set as candidate locations. The network quality degradation location estimation apparatus according to any one of claims 1 to 5.

Images