JP2011151752A - Communication network system including packet transport fault detection function - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To suppress an increase in traffic and erroneous switching frequency. <P>SOLUTION: Communication equipment positioned at a terminal point is configured to receive a packet for fault detection which is periodically transmitted from the terminal point to a fault monitoring object path, to record or update a delay fluctuation time corresponding to a difference between each reception time and the reception time of the packet received in the minimum delay time on the basis of the cycle and each reception time of the packet, to perform statistical processing on the number of received packets in each delay fluctuation time and estimate a delay fluctuation probability distribution corresponding to a probability density function using the delay fluctuation time as a probability variable, to search for a reference delay fluctuation probability as a reference for switching an object path on the basis of the estimated delay fluctuation probability distribution, to optimize the cycle and the reference delay fluctuation probability so that an utility function using both the traffic volume and erroneous switching period of the object path as variables can be made maximum, and to perform switching processing on the object path when the packet is not received within the period of a reception delay threshold determined from the reference delay fluctuating time corresponding to the optimized reference delay fluctuation probability. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a communication network system including a packet transport failure detection function.

In recent years, with the progress of Ethernet / IP technology for transporting IP (Internet Protocol) packets over Ethernet (registered trademark), communication networks are rapidly becoming IP. This trend extends to network providers, and application to backbone networks is being studied. However, high availability is required in the backbone network, and fault tolerance is an important issue.

ITU-T Recommendation Y. 1731 defines a CC (Continuity Check) function as a technique for fault management (in particular, fault detection). This CC function sets MEP (Maintenance Association End Point) at the start and end points in the intra domain, periodically transmits a CCM (Continuity Check Message) packet from one MEP, and the other MEP uses the CCM. By constantly monitoring packet communication, connection confirmation (reachability monitoring) including detection of erroneous connection between the MEPs corresponding to the start and end points is performed.

JP 2008-131198 A

IEICE technical report. IN, Information Network (IEICE technical report. Information networks 100 (208), pp.41-48, 2000.7.17), The Institute of Electronics, Information and Communication Engineers "Introduction to Analysis 2" (PP.30-37), Mitsuo Sugiura, The University of Tokyo Press, April 1985

Referring to FIG. 1, a plurality of communication devices (sometimes referred to as node devices) are provided at positions corresponding to a start point (MEP), a termination point (MEP), and a relay point (MIP: Maintenance Domain Intermediate Point). In the included packet communication network, the MEP-compliant receiving-side node device cannot receive the CCM packet within a period of 3.5 times the predetermined CCM period (see CCM Period value in FIG. 2). of continuity).

  The CC function is one of OAM (Operation Administration and Maintenance) functions. The receiving side node device reads an OpCode (Operation Code) in the OAM packet as a function identifier, and determines the type and use of the OAM packet. When the receiving side node device reads “1” in OpCode, it determines that this OAM packet is a CCM packet, and processes it as a CCM packet (see FIGS. 3, 4, and 5).

In a backbone network as a packet communication network, the service interruption time from failure detection to recovery by a redundant function is defined as 50 msec. In the ITU-T recommendation, the CCM packet transmission interval (cycle) is set to 3.33 msec, 10 msec, and 100 mse.
c, 1 sec, 10 sec, 1 min, or 10 min. In order to keep the service interruption time within 50 msec, it is necessary to use a period of 3.33 msec or 10 msec.

  When a period of 10 msec is used, LOC detection takes 35 msec, and therefore it can be spent on a path (communication path) switching process from the active path (path 1 in FIG. 1) to the backup path (path 2 in FIG. 1). The possible time is limited to 15 msec. Since it is difficult to process a large number of paths existing in the packet communication network within 15 msec, a period of 3.33 msec is used as a result. However, shortening the CCM packet cycle is inevitable that the bandwidth used for transporting the CCM packet increases.

  The length of the CCM packet is 75 bytes (see FIG. 5), and when a period of 3.33 msec is selected, a packet of about 0.2 Mbit / sec always flows between the MEP end points.

実際には、全てのパスの端点間毎にＣＣＭパケットが必要となるため、ネットワーク全体には、この整数倍のＣＣＭパケットが常時流れ続けることとなり、通常のトラフィックのために利用可能な帯域が圧迫される。

Actually, since CCM packets are required between the endpoints of all paths, this whole multiple of CCM packets will always flow through the entire network, and the available bandwidth for normal traffic will be compressed. Is done.

  In addition, when the MEP-compatible receiving side node apparatus cannot receive a CCM packet within a period that is 3.5 times the predetermined CCM cycle, it detects a LOC, that is, a packet disconnection failure. If the probability of the reception delay threshold corresponding to the period and the set value of the period are not appropriate, erroneous switching frequently occurs.

  In addition, under the condition that the service interruption time request is fixed at, for example, 50 msec, there is a trade-off relationship between the increase in traffic and the frequency of erroneous switching. Therefore, in order to satisfy both requests, the node device is appropriately cycled. And the probability serving as a reference for switching must be adjusted.

  The challenge is to provide a technology that can satisfy the service downtime requirement and optimally suppress both the increase in traffic and the increase in the frequency of erroneous switching due to failure detection packets for connection confirmation (reachability monitoring). is there.

In order to solve the above problem, a communication device located at the end point of the network that performs connection confirmation of a fault monitoring target path existing between the communication device located at the start point of the network is located at the start point Means for receiving a failure detection packet periodically transmitted from the communication device to the failure monitoring target path; received at each reception time and minimum delay time based on the cycle of the failure detection packet and each reception time; Means for recording / updating a delay variation time corresponding to a difference from the reception time of the failure detection packet; and statistically processing the number of reception of the failure detection packet for each delay variation time, thereby probing the delay variation time Means for estimating a delay variation probability distribution corresponding to a probability density function as a variable; and a criterion for switching the fault monitoring target path based on the estimated delay variation probability distribution; Means for obtaining a reference delay fluctuation probability; and means for optimizing the period and the reference delay fluctuation probability so that a utility function having both the traffic amount of the fault monitoring target path and the erroneous switching period as variables is maximized. And means for switching the fault monitoring target path when the fault detection packet is not received within a reception delay threshold value determined from a reference delay fluctuation time corresponding to the optimized reference delay fluctuation probability. Is provided.

  According to the disclosed communication device, the optimal trade-off adjustment function between the increase in traffic volume due to the failure detection packet and the increase in the erroneous switching frequency is effective for a network that handles packet transport.

  Other objects, features and advantages will become apparent upon reading the detailed description set forth below when taken in conjunction with the drawings and the appended claims.

The figure for demonstrating failure generation, LOC detection, and path switching. The figure for demonstrating the flag and period of a CCM packet. The figure which shows the frame format of an OAM packet. The figure which shows the definition of OpCode. The figure which shows the frame format of a CCM packet. 1 is a block diagram showing a schematic configuration of a communication network system according to an embodiment. The block diagram which shows the structure of a transmission side node apparatus. The block diagram which shows the structure of a receiving side node apparatus. The block diagram which shows the structure of a relay node apparatus. Inter-node device process flowchart (notifies only the transmitting side node device of an environmental change). Inter-node device processing flowchart (notifies only the receiving side node device of environmental changes). Inter-node device processing flowchart (notifying the transmission / reception node device of environmental changes). The flowchart of the failure recovery process of a receiving side node apparatus. The figure which shows the time-sequential process in a receiving side node apparatus. Correspondence table between delay variation time and number of packets. The flowchart of a statistical process. The figure which shows the relationship between delay fluctuation probability distribution and a probability. The figure which shows the classification | category by the packet loss of failure recovery time. The correlation diagram of a setting parameter and an effect. The figure which shows delay variation time and the number of packets. The figure which shows failure recovery time. Example of measurement data table (updated for each packet). Example of measurement data table (updated for each packet). Example of measurement data table (updated for each packet). Example of measurement data table (updated collectively). Example of measurement data table (updated collectively). Example of measurement data table (updated collectively).

  Hereinafter, further detailed description will be given with reference to the accompanying drawings. The drawings show preferred embodiments. However, it can be implemented in many different forms and should not be construed as limited to the embodiments set forth herein.

[System Overview]
Referring to FIG. 6 showing a system configuration in an embodiment, a communication network system SYS as a backbone network system corresponds to a start point (MEP), a end point (MEP), and a relay point (MIP) in an intra domain. A plurality of communication devices (node devices) ND are included in the position.

  This communication network system SYS is an ITU-T recommendation Y.M. In order to perform a packet transport fault detection function based on 1731, the CC function is basically used. Therefore, the CCM packet as the failure detection packet is periodically transmitted from the node device (transmission side node device) ND at the start point, and the node device (terminal device) at the end point via the node device (relay node device) ND at the relay point. By constantly monitoring the communication of the CCM packet at the receiving side node device) ND, the connection confirmation (reachability monitoring) of the failure monitoring target path between the node devices ND corresponding to the start point and the end point is performed. Here, only the path P1 and P2 as the fault monitoring target operation path and backup path are illustrated. The packet transport includes packet exchange, transmission, transmission and transfer.

  In this communication network system SYS, a CCM packet is used as a failure detection packet. However, a packet in which a bit is newly defined may be used. When using the empty bits of the CCM packet, select the appropriate empty bits from the reserved area Reserved which is initially set to “0” from 71 bytes to 74 bytes in the frame format of the CCM packet shown in FIG. Define.

  More specifically, in this communication network system SYS, the receiving side node device ND receives the failure detection packet based on the expected reception interval of the failure detection packet, that is, the transmission interval (cycle) of the failure detection packet and the actual reception interval. The delay variation is obtained, and the distribution of the reception delay is obtained. The receiving-side node device ND sets a reference delay variation time (reference delay variation time) from this distribution, and determines that a failure state occurs when a failure detection packet does not arrive within the reception delay threshold. Further, the reference delay variation time is changed by updating the reception delay distribution.

  In the communication network system SYS, the transmission side node device ND sets the transmission interval of the failure detection packet to a relatively short cycle in the initial cycle in order to increase the parameter for distribution calculation during the initial operation. After the reference delay variation time is determined by the receiving side node device ND, the transmission interval of the failure detection packet is changed to a long cycle.

  In this communication network system SYS, the erroneous switching frequency (period) is obtained from the probability corresponding to the reference delay variation time and the period, and the erroneous switching is suppressed by adjusting the period and the reference switching probability. be able to. Furthermore, the request for the failure recovery time can be satisfied, and the request for suppressing the increase in traffic volume and the increase in the frequency of erroneous switching can be simultaneously and optimally satisfied.

The main terms used in the communication network system SYS of the embodiment are defined as follows. These terms can be further understood with reference to FIG. 5, FIG. 20, FIG.
(1) Failure detection packet: A packet that is periodically transmitted to detect the occurrence of a failure in a monitored path, that is, for connection confirmation (reachability monitoring). It corresponds to the CCM packet of the OAM (operation, management, maintenance) function in the ITU-T recommendation.
(2) Failure recovery processing time: Time from failure detection until actual recovery processing is completed.
(3) Failure recovery time: The sum of the standby time from the normal reception time of the failure detection packet transmitted before the failure detection time to the failure recovery processing time.
(4) Delay time: difference between the reception time and the transmission time of the failure detection packet.
(5) Minimum delay time: the difference between the reception time and the transmission time of the failure detection packet received with the minimum delay.
(6) Delay variation time: difference between delay time and minimum delay time. That is, the difference between each reception time and the reception time of the failure detection packet that arrived with the minimum delay time.
(7) Delay variation probability distribution: probability density function with delay variation time as a random variable.
(8) Reception delay threshold: a waiting time from the normal reception time of the failure detection packet transmitted immediately before until the failure is detected in the reception side node device. Difference between failure recovery time and failure recovery processing time.

[Node device configuration]
The transmission-side node device ND, the reception-side node device ND, and the relay node device ND, which are included in the communication network system SYS illustrated in FIG. 6 and are arranged at positions corresponding to the start point, the end point, and the relay point in the intra domain, In order to perform the packet transport fault detection function, the configurations shown in FIGS. 7, 8, and 9 are employed.

  As illustrated in FIG. 7, the transmission-side node device 10 includes an environment change reception unit 11, an optimum period reception unit 12, a period setting unit 13, and a packet transmission unit 14.

  In the configuration of the communication network system SYS shown in FIG. 6, the environment change reception unit 11 of the transmission side node device 10 interfaces with the relay node device 40, and the optimum period reception unit 12 and the packet transmission unit 14 pass through the relay node device 40. And interface with the receiving side node device 20.

  As shown in FIG. 8, the receiving side node device 20 includes a packet measuring unit 21, a packet counting unit 23, a statistical processing database unit 24, a statistical processing unit 25, an optimization unit 26, a reception delay threshold setting unit 27, and a restoration process. Unit 28, optimum cycle notification unit 29, environment change reception unit 30, network information unit 31, and user interface unit UI.

  In the receiving side node device 20, the packet counting unit 23 includes a measurement data table unit 231 and a delay variation / packet number table unit 232, and the statistical processing database unit 24 includes a model unit 241. The optimization unit 26 includes a utility function calculation unit 261, a traffic calculation unit 262, and an erroneous switching period calculation unit 263.

  In the configuration of the communication network system SYS shown in FIG. 6, the packet measurement unit 21 of the reception side node device 20 interfaces with the transmission side node device 10 via the relay node device 40, and the optimum period notification unit 29 is the relay node device. The environment change receiving unit 30 interfaces with the relay node device 40 through the interface 40 with the transmission side node device 10.

  As shown in FIG. 9, the relay node device 40 includes a failure detection packet queue unit 41, a failure detection packet address reading unit 42, a traffic measurement unit 43, a traffic increase / decrease determination unit 44, an address database unit 45, and an environment change notification. A portion 46 is provided. The address database unit 45 includes a failure detection packet transmission side address 451 and a failure detection packet reception side address 452.

  In the configuration of the communication network system SYS shown in FIG. 6, the failure detection packet queue unit 41 of the relay node device 40 interfaces with the transmission side node device 10 via another relay node device 40 or directly, and notifies an environmental change notification. The unit 46 interfaces with the transmitting side node device 10 and the receiving side node device 20 via another relay node device 40 or directly.

  The transmission-side node device 10, the reception-side node device 20, and the relay node device 40 can each be configured by a computer including a processor and a memory.

[System Operation]
Next, the packet transport failure detection operation in the communication network system SYS according to the embodiment will be described with reference to FIGS. 6 to 9 and related drawings.

  10, FIG. 11 and FIG. 12 show processing procedures in the transmission side node device 10, the reception side node device 20, and the relay node device 40 during the packet transport failure detection operation. Here, in the processing procedures shown in FIGS. 10, 11 and 12, the environmental change state is changed from the relay node device 40 to one of the transmission side node device 10 and the reception side node device 20 when a sudden environmental change occurs. There are different parts depending on whether or not both are notified. This different part will be described later in the processing of [Rapid environmental change], and the processing of the common part will be described first.

[Initial cycle setting] (Process 101 in FIGS. 10, 11, and 12)
An operator (administrator) of the transmission side node device 10 (see FIG. 7) arranged at a position corresponding to the starting point in the communication network system SYS shown in FIG. Start the failure detection operation.

  In response to this, the packet transmission unit 14 in the transmission side node device 10 is for detecting a failure toward the reception side node device 20 via the plurality of relay node devices 40 in accordance with an initial period predetermined by the period setting unit 13. Send packets periodically. In the following description, passing through the relay node device 40 may be omitted.

  In order to satisfy the failure recovery request, it is necessary to quickly detect the failure, and it is necessary to transmit at a sufficiently small cycle (for example, 3.33 msec) as an initial cycle. Sufficiency of this initial period can be judged statistically, but there is no clear ground when statistical data is not sufficient. However, since the traffic generally increases as the period becomes smaller, the operator of the transmitting side node device 10 is required to carefully determine the initial period so that the period is not too small.

[Initial threshold setting] (Processing 102 in FIGS. 10, 11, and 12)
The operator in the receiving-side node device 20 (see FIG. 8) sets the initial threshold for failure recovery to a period that is long enough to suppress the frequency of erroneous switching. The initial threshold is (1) a value obtained by subtracting the initial period and the failure recovery processing time from the fault recovery time 50 msec, that is, the maximum value that can be set as the initial threshold, (2) an appropriate constant value that does not depend on the initial period, ( 3) As in the ITU-T recommendation, the value is 3.5 times the initial period (for example, 3.33 × 3.5 = 11.7 msec when the initial period is 3.33 msec. If not, there is no clear basis for the appropriateness of the initial threshold.

  Until the reception side node device 20 derives the reference delay variation time according to the subsequent processing and notifies the transmission side node device 10, the reception side node device 20 sets this initial threshold value as the reference delay variation time, and detects the failure. When the packet does not arrive within the initial threshold period (when the delay exceeds the threshold value), it is determined that a failure has occurred, and failure recovery processing including path switching is performed (processing 102, 131, 132, 133 in FIG. 13). ).

  FIG. 14 shows time-series processing until the receiving-side node device 20 normally receives packets 1 to 5 among the failure detection packets and detects a failure in the packet 6. The receiving side node device 20 waits from the normal reception of the packet 5 to the reception delay threshold, and starts the failure recovery processing at the time when the reception delay threshold is exceeded. The reception times from packet 1 to packet 5 are recorded in the measurement data table unit 231.

[Delayed data accumulation] (Process 103 in FIGS. 10, 11, and 12)
In the reception-side node device 20, each time the packet measurement unit 21 receives a failure detection packet periodically transmitted from the transmission-side node device 10, the packet counting unit 23 records the reception time in the measurement data table unit 231. To do. The packet counting unit 23 updates the delay variation time as needed based on the reception time and period each time one measurement data is recorded in the measurement data table unit 231 or each time a plurality of measurement data is recorded. A specific example of the method for updating the measurement data table unit 231 will be described later in the “Update of measurement data table unit” process.

  Next, the packet counting unit 23 totals the number of packets for each delay variation time from the measurement data table unit 231, and a correspondence table between the delay variation time and the packet number as shown in FIG. Create as H.232. The sufficiency of the data amount is determined quantitatively depending on whether or not the suitability for the data of the delay variation probability distribution obtained as a result of statistical processing to be described later is sufficient for the request for the fit.

[Statistical processing] (Process 104 in FIGS. 10, 11 and 12)
In the receiving side node device 20, the statistical processing unit 25 statistically processes the delay variation / packet number table unit 232 to estimate the delay variation probability distribution. Here, in one-way communication, it is important that the receiving side node device 20 knows the period and can accurately know the reception time by increasing the accuracy of the internal clock. With these two conditions, it is possible to accurately obtain the delay variation probability distribution and the probability / statistics.

  Next, details of this statistical processing will be described based on the statistical processing procedure shown in FIG.

<Network information confirmation> (process 1041 in FIG. 16)
The statistical processing unit 25 in the receiving-side node device 20 determines the network topology of each node device constituting the communication network system SYS, the specifications of each node device (for example, the start point, the end point, the relay point, etc.), the average traffic amount, and the like. The included network information is acquired from the database of the network information unit 31.

  This network information may be set manually by each node device operator such as the receiving side node device 20 through the user interface unit UI, or may be automatically collected by each node device. When an abrupt environmental change occurs, information notified by a “rapid environmental change” process described later is held in the network information unit 31. However, when a node device does not hold network information as an initial state, such as when a new node device is added, a notification-dedicated protocol, packet, bit, etc. are appropriately defined and used. Specifically, the packet, bit, and usage are newly defined in the OAM packet.

<Select traffic model> (Process 1042 in FIG. 16)
The statistical processing unit 25 obtains a traffic model (probability density function) that is a candidate for the delay variation probability distribution of each path from the model unit 241 of the statistical processing database unit 24 based on the network information obtained in the network information confirmation processing 1041. select.

  The model unit 241 is added and updated irregularly with changes and developments in traffic theory and model theory. For example, in an IP communication network that passes through a plurality of node devices, the Pareto distribution is the optimal model when the network is congested, and the lognormal distribution is the optimal model when the network is free (see Non-Patent Document 1 for details).

<Fitting of Model and Data> (Process 1043 in FIG. 16)
Next, the statistical processing unit 25 fits each model and data (network information). A parameter estimation method (maximum likelihood method or the like) is used as a method for fitting a traffic model to data. For example, when the traffic model has an exponential distribution, the probability density function g (x) is given by the following equation using the parameter λ ≧ 0. Here, by obtaining the parameter λ by parameter estimation, a probability density function g (x) to be fitted to the data is obtained.

統計処理部２５は上述した処理１０４２，１０４３をトラフィックモデル候補の全てについて繰り返す（図１６中の処理１０４４）。

The statistical processing unit 25 repeats the above-described processes 1042 and 1043 for all the traffic model candidates (process 1044 in FIG. 16).

<Compatibility comparison> (Processing 1045 in FIG. 16)
The statistical processing unit 25 compares the goodness of fit between the probability density function as a candidate and the data using a chi-square x ² test (Chi-square Test).

<Delay variation probability distribution determination> (Process 1046 in FIG. 16)
The statistical processing unit 25 selects the traffic model with the highest fitness as the optimal traffic model, and uses the probability density function with the highest fitness as the delay variation probability distribution in the subsequent processing.

<Determination of reference delay variation time> (Process 1047 in FIG. 16)
The statistical processing unit 25 uses the delay variation probability distribution obtained in the delay variation probability distribution determination processing 1046 and uses a probability (reference delay variation probability: 10 ⁻⁷ percent, for example) as a reference for switching necessary for determining the reception delay threshold. ) Corresponding to the reference delay variation time (reference delay variation time).

  Instead of using the probability as a reference, a statistic such as a standard deviation corresponding to the probability may be used. However, the standard cannot be generally set as “integer multiple of standard deviation”. This is because when the delay variation probability distribution is a normal distribution, the probability that a random variable is included in a range where the deviation from the average is ± 1σ or less is about 68.3%, and if it is ± 2σ or less, about 95.4%. If it is ± 3σ, it is about 99.7%, but a general continuous distribution that is not a normal distribution cannot be simply stated in this way.

Referring to FIG. 17, the probability that a delay variation of t = T1 or more occurs is an area S of a portion surrounded by a delay variation time axis, a line indicating time t = T1, and a delay variation probability distribution line. The procedure for obtaining the reference delay variation time from the reference delay variation probability is as follows.
(1) The area S, that is, the integral value is obtained by numerical calculation or the like.
(2) The probability value for each change is calculated while changing T1.
(3) The delay variation time when it coincides with the reference delay variation probability is set as the reference delay variation time.
The method for determining the reference delay variation probability will be described later in the [optimal period / probability determination] process.

  Here, it is assumed that the probability of packet loss of a failure detection packet (for example, a CCM packet) is very small. When packet loss cannot be ignored, the probability that consecutive packet losses occur is extremely small, and the reception delay threshold is set to be an integer multiple of the period larger than when no packet loss occurs (can be ignored) (see FIG. 18). .

To show this, assuming that the probability of packet loss is P, if the event that causes packet loss is independent, the probability of packet loss occurring n (n ≧ 2) times consecutively is P ⁿ . As n increases, P ⁿ decreases exponentially. If P is sufficiently small, P ² is negligibly small, so it is sufficient to increase the reception delay threshold by one period.

The reference delay fluctuation time for failure recovery can also be obtained from the probability density function focusing on the arrival interval of the failure detection packet. This probability density function can be calculated using the delay variation probability distribution f (s) for one packet. Here, s is the difference between the reception time and the arrival time when there is no delay variation. If the reception time of the previous failure detection packet is t ₁ , the reception time of the next failure detection packet is t ₂ , the period is T, and the difference between the reception times of failure detection packets is t = t ₂ −t ₁ , The delay variation probability distribution F (t) of the arrival time difference is expressed by convolution of the difference as shown in the following equation using the delay variation probability distribution f (s) for one packet. When convolution is used, the amount of calculation and the error due to the calculation become large.

到着間隔の確率分布モデルの候補を統計処理用データベース部２４のモデル部２４１に保持することによって、上記処理１０４１〜１０４７の手順で遅延変動確率分布を直接求めることもできる。ただし、到着時間差の確率分布モデルは一般的によく知られているパレート分布、指数分布、対数正規分布、ガンマ分布などの連続確率分布にはならない。また、この遅延変動確率分布は１つ前に送信された障害検出用パケットの受信側ノード装置２０における遅延変動と、１つ後に送信された障害検出用パケットの受信側ノード装置２０における遅延変動とを両方考慮した分布になるので、１つ前の障害検出用パケットの受信時刻が確定した後は、この分布を用いないほうが正確である。

By holding the probability distribution model candidates for the arrival interval in the model unit 241 of the statistical processing database unit 24, the delay variation probability distribution can be directly obtained by the procedure of the above processings 1041 to 1047. However, the probability distribution model of the arrival time difference does not become a continuous probability distribution such as a generally well-known Pareto distribution, exponential distribution, lognormal distribution, or gamma distribution. Also, this delay variation probability distribution is the delay variation in the failure detection packet transmitted in the previous one in the reception side node device 20 and the delay variation in the failure detection packet transmitted in the next transmission in the reception side node device 20. Therefore, it is more accurate not to use this distribution after the reception time of the previous failure detection packet is determined.

[Optimum period / probability determination] (Process 105 in FIGS. 10, 11, and 12)
FIG. 19 shows the correlation between the traffic amount of the failure detection packet, the erroneous switching period, and the failure recovery time. Since the failure recovery processing time is an eigenvalue for each node device, the failure recovery time varies depending on the period dependent portion and the reference delay variation time. When the packet loss is taken into consideration, the period dependent portion is a value obtained by subtracting the delay variation time of the failure detection packet received immediately before from an integer multiple of the period.

Regarding the relationship between the period-dependent part and the delay variation time, the receiving side node apparatus 20 transmits the optimum period notification packet to the transmitting side node apparatus 10 after normal reception of the previous failure detection packet, and is optimal. Since it is not in time to notify the period and change the period, the period cannot be changed after the arrival of the previous reception failure detection packet. However, since the reference delay variation probability (reference delay variation time), which is a set value in the receiving side node device 20, can be dynamically changed according to the reception time of the previous failure detection packet, The reference delay variation time can be changed according to the delay variation time of the failure detection packet. As another countermeasure, there is a method of ignoring the delay variation time of the previous reception failure detection packet and calculating by excluding the average value of the previous delay variation time from the failure recovery time.

  In FIG. 19, the parameters that can be set by the receiving side node device 20 are the period and the reference delay variation probability. Decreasing the period increases traffic and reduces the erroneous switching period, but reduces the failure recovery time. When the reference delay variation probability is increased, the erroneous switching period is reduced, but the failure recovery time is reduced. Using this correlation, the operator of the receiving side node device 20 adjusts these values to determine the optimum period. Specifically, the optimum cycle is determined by the following procedure.

<Calculation of erroneous switching period>
The erroneous switching period calculation unit 263 of the optimization unit 26 in the receiving side node device 20 calculates the erroneous switching period. If the reference delay variation probability is P ₁ and the period of the failure detection packet is τ,
The expected value of the erroneous switching occurrence period (erroneous switching period) can be calculated as τ / P ₁ . For example, when the reference delay variation probability is set to 10 ⁻⁷ percent and the failure detection packet is transmitted at a cycle of 10 msec, the erroneous switching period can be calculated as 3.2 [year] from the following equation.

ここで、例として挙げた１０^-7パーセントの妥当性であるが、一般的に確率分布が正規分布に従う場合、６σ（標準偏差σ）だけ平均値より外れる確率のオーダーは、統計学では１０億分の１、すなわち１０^-7パーセントである。一般の連続確率分布でも、平均値から標準偏差の整数倍離れるだけで確率は急激に減少する。なお、ゼロへの収束速度は確率分布により異なる。

Here, the validity of 10 ⁻⁷ percent given as an example, but in general, when the probability distribution follows a normal distribution, the order of the probability that deviates from the average value by 6σ (standard deviation σ) is 1 billion in statistics. A fraction, or 10 ^-7 percent. Even in a general continuous probability distribution, the probability decreases rapidly only by moving away from the average value by an integral multiple of the standard deviation. Note that the convergence speed to zero varies depending on the probability distribution.

<Calculation of traffic volume>
If the period is τ and the failure detection packet length is L, the traffic volume is L / τ. For example, the CCM packet length is 75 bytes (see FIG. 5), and when a period of 3.33 msec is selected, the traffic volume can be calculated as follows. The traffic calculation unit 262 calculates the traffic amount by the following formula.

〈効用関数の計算〉
周期と基準遅延変動確率との最適化は、定量的には、次の方法を用いる。受信側ノード装置２０の運用者はトラフィック量ξと誤切り替え期間ζとを変数とする効用関数（目的関数）ｌ（ξ，ζ）を定義する。この効用関数とは、トラフィック量と誤切り替え期間とによって運用者が得る効用、あるいはトラフィック量と誤切り替え期間との要求に対する満足度を表す関数のことである。トラフィック量は小さいほど満足度が高く、誤切り替え期間は大きいほど満足度が高いので、ζ−ξは効用関数の例である。

<Calculation of utility function>
For the optimization of the period and the reference delay variation probability, the following method is used quantitatively. The operator of the receiving side node device 20 defines a utility function (objective function) l (ξ, ζ) having the traffic amount ξ and the erroneous switching period ζ as variables. The utility function is a function representing utility obtained by the operator based on the traffic volume and the erroneous switching period, or a satisfaction degree with respect to a request for the traffic volume and the erroneous switching period. The smaller the traffic volume, the higher the satisfaction level, and the longer the erroneous switching period, the higher the satisfaction level.

  When the request for the failure recovery time η is expressed by an equation, the utility function calculation unit 261 uses, for example, η−50 = 0 (50 represents 50 msec) as a constraint, and Lagrange's undetermined multiplier method (details are non-patent). The period and the reference delay variation probability are adjusted so as to maximize the utility function l (ξ, ζ) using the reference 2). The utility function calculation unit 261 uses linear programming (see Non-Patent Document 3 for details) when the constraint condition is expressed by an inequality, for example, ζ ≧ 5 [year], ξ ≦ 0.20 [Mbit / sec]. ) To maximize the utility function. The period when the utility function is maximized and the reference delay variation probability are optimum set values.

  Using the above method, optimize the amount of traffic due to failure detection packets in a trade-off relationship and the erroneous switching period by adjusting the cycle of failure detection packets and the probability of switching. be able to.

  Further, when the reference delay variation time is sufficiently small, the period of the failure detection packet can be increased. For example, in the case of a CCM packet, even if a CCM period larger than 3.33 msec, for example, 10 msec is used, a failure recovery time of 50 msec can be realized. In general, as the period increases, the amount of traffic decreases, and thus the increase in traffic due to the CCM packet can be suppressed in this way.

  Furthermore, if the erroneous switching period determined by the ratio of the period and the reference delay variation probability is sufficiently large, that is, if the reference delay variation probability is sufficiently small compared to the period of the failure detection packet, the frequency of erroneous switching is suppressed as required. Can do.

[Reception delay threshold setting] (Process 106 in FIGS. 10, 11, and 12)
The optimization unit 26 sets a reception delay threshold value determined from the reference delay variation time corresponding to the optimized reference delay variation probability in the reception delay threshold setting unit 27. The reception delay threshold setting unit 27 determines that a failure has occurred according to the set reception delay threshold when the failure detection packet does not arrive within the period of the reception delay threshold, and a failure recovery process including path switching by the recovery processing unit 28 To do.

  The transmission-side node device 10, the reception-side node device 20, and the relay node device 40 that comply with the ITU-T recommendation, depending on the value (value) written in the OpCode of the OAM packet, Is determined (see FIGS. 3 and 4). For example, when “1” is recorded in the OpCode value, it is determined that the packet is a CCM packet used for the CC function, and the CCM packet is processed. Hereinafter, when referring to a packet other than a failure detection packet, the reserved OpCode value is used as shown in FIG. 4 and the defined OAM PDU (Protocol Data Unit) type is associated with the value. And

[Notification of Optimal Period] (Process 107 in FIGS. 10, 11, and 12)
The optimum cycle notification unit 29 notifies the sending side node device 10 of the optimum cycle. In order to notify the optimum cycle, a cycle notification packet is transmitted from the reception-side node device 20 to the transmission-side node device 10. For the periodic notification packet, a new periodic notification packet is defined and used as the OAM packet. In order to express the cycle, a bit for writing the cycle is defined in the cycle notification packet, and the cycle is dynamically written here. For example, from 1 msec to 50 msec is a number 1 to 50 in milliseconds and is explicitly expressed by 6 bits. Further, when using the cycle of the CC function recommended by ITU-T, for example, notification is made to change the cycle from 3.33 msec to 10 msec.

[Optimum cycle setting] (Process 108 in FIGS. 10, 11, and 12)
In the transmitting side node device 10, the optimum cycle receiving unit 12 receives the cycle notification packet transmitted from the receiving side node device 20, reads out the bit in which the optimum cycle is written from the cycle notification packet, and reads the optimum The period is set in the period setting unit 13. The packet transmission unit 14 transmits a failure detection packet at the set optimum period. As a result, the receiving side node device 20 receives the failure detection packet at the optimum cycle and repeats the above-described processing.

[Abrupt environmental change] (Process 201 in FIGS. 10, 11, and 12)
In the communication network system SYS, the network configuration has changed due to a newly added relay node device 40, a deleted relay node device 40, or an increase or decrease in the average traffic volume. When this occurs, the delay variation probability distribution in the receiving side node device 20 needs to be updated, so it is necessary to re-acquire data necessary for the parameter for calculating the delay variation probability distribution.

  A node device having a network monitoring function, for example, the relay node device 40 constantly monitors a rapid environmental change that affects the delay variation probability distribution by the following method. When a sudden environmental change event occurs, the relay node device 40 sets an environmental change notification packet with a bit indicating that there is a sudden environmental change in at least one of the transmission-side node device 10 and the reception-side node device 20. Send to notify.

  This environment change notification packet uses a newly defined OpCode. For example, “6” is used as the OpCode value, and ENM (Environment Notification Message) is defined as the OAM PDU type to correspond to it. Further, in order to notify what is a sudden change event, an appropriate bit is defined and used in the frame format of the environment change notification packet. For example, a bit string of bit length 3 is defined, and “001” corresponds to increase / decrease in average traffic volume, “010” corresponds to addition of a new node, and the like. Also, as the degree of environmental change, the traffic increase / decrease amount is separately defined and notified. A specific processing procedure is shown below.

<Change in average traffic volume>
The function of the relay node device 40 in monitoring traffic increase / decrease will be described with reference to FIG. The traffic measurement unit 43 in the relay node device 40 counts the traffic amount of a queue (failure detection packet queue unit) 41 in which failure detection packets passing through the own device are temporarily accumulated.

The traffic increase / decrease determination unit 44 determines whether or not it can be said that the traffic amount has increased or decreased statistically based on the traffic volume counting result by the traffic measurement unit 43.
(1) An appropriate traffic calculation period for comparing the average traffic volume is determined.
(2) The traffic average amount is calculated in time series in the predetermined traffic calculation period. (3) A test is used for the null hypothesis that the average traffic volume is constant.

  When the null hypothesis is rejected, the traffic increase / decrease determination unit 44 reads all the failures that the failure detection packet address reading unit 42 has read from the failure detection packet queue unit 41 and recorded in the address database unit 45 in advance. For environment change notification with a bit indicating that there has been an environment change for at least one of the transmission side node device address (transmission side address) 451 and the reception side node device address (reception side address) 452 of the detection packet. The packet is transmitted from the environment change notification unit 46. Here, the transmission side node device address 451 and the reception side node device address 452 correspond to the identifier MEPID (see FIG. 5) in the CC function of the ITU-T recommendation.

  When there are a plurality of queues used for buffering failure detection packets in the failure detection packet queue unit 41, the traffic volume is monitored for each queue. For the sudden change event, in the above example, “001” is a bit representing the occurrence of an increase or decrease in the average traffic volume, so this bit is set. As for the degree of environmental change, for example, a bit having a bit length of 10 is defined and the unit is used as Mbit / sec in order to notify the increase / decrease amount of the average traffic. For example, if an average traffic decrease of 10 Mbit / sec occurs, the bit is “1000001010” (where the first bit “1” indicates a traffic decrease and “0” indicates a traffic increase in units of Mbit / sec. ) Is sent.

<For node addition / deletion>
When the relay node device 40 is newly added, in the added relay node device 40, the failure detection packet address reading unit 42 transmits the node device on the transmission side written in the failure detection packet passing through the own device. The address and the receiving-side node device address are read from the failure detection packet queue unit 41 and recorded in the address database unit 45. The relay node device 40 transmits an environment change notification packet in which a bit indicating that a new node has been added as a sudden environmental change (in the above example, “010” with a bit length of 3 bits) is set. It transmits to at least one of the address 451 and the receiving side node apparatus address 452.

  When the relay node device 40 is deleted, in the relay node device 40 to be deleted, a bit indicating that there is a node deletion as an abrupt environmental change (in the above example, a bit having a bit length of 3 is “100”). Is transmitted to at least one of the transmission-side node device address 451 and the reception-side node device address 452. Instead of this processing, a bit indicating that the node has been deleted as a sudden environmental change in the relay node device 40 adjacent to the deleted relay node device 40 (in the above example, a bit having a bit length of 3). The environment change notification packet with “100”) may be transmitted to at least one of the transmission-side node device address 451 and the reception-side node device address 452.

<Notice of sudden environmental change>
Subsequently, the notification destination of a rapid environmental change is divided into three cases, and the processing is individually described for each. In FIG. 10, FIG. 11 and FIG. 12, the flow corresponding to the rapid environmental change is indicated by a thick solid line.

(1) When notifying only the transmitting side node device 10:
When a sudden environmental change occurs (process 201 in FIG. 10), the relay node device 40 transmits an environment change notification packet from the environment change notification unit 46 only to the transmitting side node device 10. In the transmission side node device 10, when the environment change receiving unit 11 receives the environment change notification packet, the cycle setting unit 13 sets the cycle to the initial cycle as the environment change initial process (process 202 in FIG. 10). The receiving side node device 20 is notified by the following two methods.

  (1-1) A failure detection packet with a bit indicating data recovery is transmitted from the packet sending unit 14 to the receiving-side node device 20. For example, one bit of the 72nd byte in the frame format of the CCM packet shown in FIG.

  (1-2) A special cycle (recovery cycle: for example, 1 msec) used only for data re-definition is defined in advance, and a failure detection packet with a bit indicating this cycle is set from the packet sending unit 14 to the receiving-side node device 20 Send to. For the definition of the re-recovery period, for example, 4 to 7 bits of the Reserved bits of the 3rd Flag in the frame format of the CCM packet shown in FIG. 5 are used. Then, the cycle bit is expanded and used, for example, “10000”, and the recovery cycle (1 msec) is associated with this bit. When receiving the notification of the sudden change in environment from the transmitting side node device 10, the receiving side node device 20 changes the reception delay threshold value to the initial threshold value and performs data re-acquisition. Thereafter, the normal flow related to the processing of the common part described above is followed.

(2) When notifying only the receiving side node device 20:
The relay node device 40 transmits an environment change notification packet from the environment change notification unit 46 only to the receiving side node device 20 when a sudden environment change occurs (process 201 in FIG. 11).

  (2-1) The receiving side node device 20 that has received the environment change notification packet by the environment change receiving unit 30 performs an environment change initial process (process 203 in FIG. 11). In other words, the initial period is set as the optimum period from the optimum period notifying unit 29 and the period notification packet is transmitted to the transmitting side node device 10. In the transmission side node device 10, when the optimum period reception unit 12 receives the period notification packet, the transmission side node device 10 sends a failure detection packet from the packet transmission unit 14 to the reception side node device 20 in the initial period set in the period setting unit 13. Send.

(2-2) Further, in the receiving side node device 20 that has received the environment change notification packet by the environment change receiving unit 30, the initial threshold value setting (in FIG. Processing 102) is performed, and data re-acquisition is started. After that, follow the normal flow. The transmitting side node device 10 and the receiving side node device 20 perform these processes (2-1) and (2-2) in parallel.

(3) When notifying both the transmitting-side node device 10 and the receiving-side node device 20:
The relay node device 40 transmits an environment change notification packet from the environment change notification unit 46 to both the transmission-side node device 10 and the reception-side node device 20 when an abrupt environment change occurs (process 201 in FIG. 12). To do.

  (3-1) In the transmission-side node device 10 that has received the environment change notification packet by the environment change receiving unit 11, the packet sending unit 14 performs the period setting unit 13 as an environment change initial process (process 202 in FIG. 12). The failure detection packet is periodically transmitted in the initial period set in (1).

  (3-2) In the receiving side node device 20 that has received the environment change notification packet by the environment change receiving unit 30, initial threshold setting (process 102 in FIG. 12) is performed as the environment change initial process (process 203 in FIG. 12). ) To start data recovery. After that, follow the normal flow. The transmission side node device 10 and the reception side node device 20 perform these processes (3-1) and (3-2) in parallel. In the case of this process (3), there is an advantage that a time lag does not occur in the process between the transmission side node apparatus 10 and the reception side node apparatus 20 as compared with the processes (1) and (2).

[Update measurement data table]
The case where the delay variation time of the measurement data table unit 231 is updated every time one packet newly arrives at the receiving side node device 20 will be described with reference to FIGS. 22, 23, and 24.

  In the update processing of the delay variation time, information on the old packet (here, packet 1) is not discarded. It is assumed that 10 msec is set as the period of the failure detection packet. For the sake of simplicity, the time at which packet 1 arrives (reception time) is set to “.000000000000 (only comma seconds or less)”. A circle ○ in the minimum delay time column indicates a packet candidate (minimum delay) in which the corresponding packet arrives at the delay time closest to the minimum delay time (see FIG. 20), that is, a packet that arrives at the receiving side node device 20 at the minimum delay time. Packet).

  First, referring to FIG. 22, the content of the measurement data table represents a state in which packets 1 to 6 are received by the receiving side node device 20. Here, the comparison time is a time to be compared with the reception time for calculating the delay variation time, and the delay variation time is a difference between the reception time and the comparison time. Since the reference of the minimum delay time is the packet 1, the reference time of the comparison time is the reception time “.000000000000” of the packet 1, and the cycle is an interval of 10 msec. Therefore, the comparison time from the packet 2 to the packet 6 is “. "01" to ".05" are entered (held). The delay variation for each packet is represented by the difference between the reception time of each packet and the comparison time.

Next, the measurement data table shown in FIG. 23 will be described. When the reception side node device 20 receives the packet 7, the difference between the reception time of the packet 7 and the comparison time becomes negative. Therefore, the candidate of the minimum delay packet is the packet 7, and the circle ○ in the minimum delay time column Is changed from packet 1 to packet 7. Since the reference packet of the comparison time is the packet 7, the delay variation time of the packet 7 becomes 0, and “0.006 (μsec)” which is the difference between the comparison time based on the packet 1 and the comparison time based on the packet 7 is set. Each is added to the delay variation time from packet 1 to packet 6. This delay variation time is the delay variation time for the packets 1 to 7.

  Next, the measurement data table shown in FIG. 24 will be described. When the receiving side node device 20 receives the packet 8, the comparison time of the packet 8 is set such that the reception time of the packet 7 is the reference time. 059999994+. 010 (cycle) =. 069999994. When the difference between the reception time of packet 8 and the comparison time is calculated,. 0699999990-. Since 069999994 = −0.004 (μs), the packet 8 becomes the reference time of the minimum delay time. The delay variation time of the packet 8 becomes “0”, and “0.004 (μsec)”, which is the difference between the comparison time based on the packet 7 and the comparison time based on the packet 8, is added to the packet 1 from the packet 1 respectively. The delay variation time obtained by this operation is the delay variation time for the packets 1 to 8.

  Next, a case where the delay variation time of the measurement data table unit 231 is updated at once after the reception side node device 20 receives a plurality of packets will be described with reference to FIGS. 25, 26, and 27.

  In this delay variation time update process, the cycle is 10 msec, and the measurement data table is updated when data of eight packets is accumulated. For the sake of simplicity, the time when packet 1 arrives is set to “.000000000000 (only for comma seconds or less)”.

  First, referring to FIG. 25, this measurement data table is the same as the measurement data table shown in FIG. 22, but differs in that a circle ○ is not written in the minimum delay time column. This is because the candidate for the minimum delay packet is not determined until eight packets are received. This minimum delay time column is filled in after all the packets necessary for updating the delay variation time are received.

  Next, referring to FIG. 26, the content of the measurement data table indicates a state when the packets 7 and 8 further arrive. Here, as the delay variation time, the difference between the reception time of packet 7 and packet 8 and the comparison time is written. Negative values are entered as they are.

  Next, referring to the measurement data table shown in FIG. 27, since “−0.010 (μsec)” of packet 8 is the minimum value in the delay variation time, the reception time of packet 8 is the reference time of the minimum delay time. Thus, a circle ○ is entered in the minimum delay time column of the packet 8. Subsequently, “0.010 (μsec)” is added to the delay variation time from packet 1 to packet 8 so that the delay variation of packet 8 becomes “0”. The delay variation time from packet 1 to packet 8 obtained in this way is entered in the delay variation time column.

  In the update processing of the delay variation time described above, an update method in which the data of the oldest packet is discarded each time new data for one packet is obtained, or new data of a plurality of packets is recorded. A method of discarding data and updating the data table may be employed. For example, the arrival time of a candidate for a minimum delay packet that is a reference may be held in a database different from the measurement data table unit 231 and the measurement data table may be updated using the arrival time as a reference time.

[Modification]
The processing in the above-described embodiment is provided as a computer-executable program, and can be provided via a recording medium such as a CD-ROM or a flexible disk, and further via a communication line.

  In addition, each of the processes in the above-described embodiment can be performed by selecting and combining any or all of the processes.

SYS Communication network system ND Node device P1 Operation path P2 Backup path 10 Transmission side node device 20 Reception side node device 40 Relay node device

Claims (5)

A communication device located at the end point of the network for confirming connection of a fault monitoring target path existing between the communication device located at the start point of the network;
Means for receiving a failure detection packet periodically transmitted from the communication device located at the starting point to the failure monitoring target path;
Means for recording / updating a delay variation time corresponding to a difference between each reception time and the reception time of the failure detection packet received at the minimum delay time based on the period of the failure detection packet and each reception time;
Means for statistically processing the number of received fault detection packets for each delay variation time to estimate a delay variation probability distribution corresponding to a probability density function having the delay variation time as a random variable;
Means for obtaining a reference delay fluctuation probability serving as a reference for switching the fault monitoring target path based on the estimated delay fluctuation probability distribution;
Means for optimizing the period and the reference delay variation probability so that a utility function having both the traffic amount of the fault monitoring target path and the erroneous switching period as variables becomes maximum;
Means for performing switching processing of the failure monitoring target path when the failure detection packet is not received within a reception delay threshold period determined from a reference delay variation time corresponding to the optimized reference delay variation probability;
A communication device comprising: In order to increase the parameter for estimating the delay variation probability distribution, after receiving the failure detection packet in a predetermined initial cycle, the failure detection packet is detected in the optimized cycle having a longer cycle than the initial cycle. The communication device according to claim 1, which receives a packet. The data necessary for the parameter for estimating the delay variation probability distribution is re-acquired when an environment change notification packet indicating an environment change is transmitted from a communication device located at a relay point having a network monitoring function. Communication device. The communication apparatus according to claim 1, wherein the failure detection packet is a CCM packet in which an empty bit in a reserved area is defined. A method for confirming connection of a fault monitoring target path existing between a communication device located at a network start point and a communication device located at a network end point;
Receiving a failure detection packet periodically transmitted from the communication device located at the starting point to the failure monitoring target path;
Recording and updating a delay variation time corresponding to a difference between each reception time and the reception time of the failure detection packet received at the minimum delay time based on the period of the failure detection packet and each reception time;
Statistically processing the number of received failure detection packets for each delay variation time to estimate a delay variation probability distribution corresponding to a probability density function having the delay variation time as a random variable;
Obtaining a reference delay variation probability serving as a reference for switching the failure monitoring target path based on the estimated delay variation probability distribution;
Optimizing the period and the reference delay variation probability so that the utility function having both the traffic amount of the fault monitoring target path and the erroneous switching period as variables is maximized;
When the failure detection packet is not received within a reception delay threshold period determined from a reference delay variation time corresponding to the optimized reference delay variation probability, the failure monitoring target path is switched;
A method in which a communication device located at the end point executes this.

Images