JP4937191B2 - Packet delay data processing apparatus, packet delay data processing method and program - Google Patents

Description

  The present invention relates to a packet delay data processing device, a packet delay data processing method, and a program.

Conventionally, various methods for measuring an end-to-end packet delay in an IP (Internet Protocol) network have been proposed. For example, in the techniques described in Non-Patent Documents 1 and 2, the influence of queuing delay is removed from the measured packet delay data using linear programming or piecewise minimum values, and packet delay fluctuations are extracted. is doing.
V. Paxson, “On Calibrating Measurements of Packet Transit Timer.” In Proc. ACM SIGMETRICS 1998, pp.11-21, 1998. SB Moon, P. Skelly and D. Towsley, “Estimation and Removal of Clock Skew from Network Delay Measurements.” In Proc. IEEE INFOCOM, pp.227-234, March 1999.

However, the techniques described in Non-Patent Documents 1 and 2 have a problem in that linear correction may not be applied to packet delay data due to discontinuity of packet delay data accompanying a route change or the like. Further, in order to measure the packet delay, it is necessary that the time of the packet transmission terminal and the reception terminal are synchronized. When measuring packet delay data between multiple points, the time lag of one terminal affects many paths. Therefore, if all terminals are not synchronized in time, highly accurate packet delay data can be obtained. I can't.
The present invention has been made in view of the above points, and an object of the present invention is to enable accurate correction by applying linear correction even for packet delay data having discontinuity due to a path change or the like. A packet delay data processing device, a packet delay data processing method, and a program are provided.

  The present invention has been made to solve the above problems, and one aspect of the present invention is a packet delay data processing apparatus that corrects packet delay data between a plurality of terminals distributed and arranged in a network. Storage means for storing packet delay data of each path between the terminals, decomposition means for decomposing the packet delay data stored in the storage means into a high frequency component and a low frequency component by Fourier transform or wavelet transform, and the decomposition Packet delay data processing comprising: calculating means for calculating an average value of the low frequency components for each path; and adding means for adding high frequency components to the average value to obtain corrected packet delay data Device.

  In addition, according to an aspect of the present invention, the packet delay data processing apparatus includes an optimization unit that optimizes an average value of the low frequency components of all paths by nonlinear programming so as to satisfy a predetermined constraint condition. It is characterized by that.

  According to another aspect of the present invention, in the above packet delay data processing device, the optimization unit may use an objective function defined using the degree of instability of each terminal based on a dispersion value of the packet delay data of each terminal. It is characterized by using.

  Another aspect of the present invention is a packet delay data processing method that includes storage means for storing packet delay data of a path between terminals and corrects packet delay data between a plurality of terminals distributed in a network. A step of decomposing the packet delay data stored in the storage means into a high frequency component and a low frequency component by Fourier transform or wavelet transform; a step of calculating an average value of the low frequency component; A packet delay data processing method characterized by comprising:

  In addition, according to one aspect of the present invention, the packet delay data processing method includes an optimization step of optimizing an average value of the low frequency components of all paths by nonlinear programming so as to satisfy a predetermined constraint condition. It is characterized by that.

  In addition, according to one aspect of the present invention, in the packet delay data processing method, the optimization step includes an objective function defined using the instability of each terminal based on a dispersion value of the packet delay data of each terminal. It is characterized by using.

  One embodiment of the present invention is a program for correcting packet delay data between a plurality of terminals distributed and arranged in a network, and a computer comprising storage means for storing packet delay data of a path between the terminals. The packet delay data stored in the storage means is Fourier-transformed or wavelet-transformed into a high-frequency component and a low-frequency component; an average value of the low-frequency component is calculated; And a step of adding the components to obtain corrected packet delay data.

  Further, one aspect of the present invention is characterized in that the program includes an optimization step of optimizing an average value of the low frequency components of all paths by nonlinear programming so as to satisfy a predetermined constraint condition. To do.

  The optimization step uses an objective function defined using the instability of each terminal based on a dispersion value of packet delay data of each terminal.

  According to the present invention, the packet delay data is subjected to Fourier transform or wavelet transform to be decomposed into a high frequency component and a low frequency component, and an average value of the decomposed low frequency components is calculated for each path. Since the average value of the low frequency components can be a fixed component of the packet delay, the fixed component of the packet delay can be linearly corrected. Thereby, it is possible to apply the linear correction and perform the correction with high accuracy. Further, since the high frequency component can be a variation in packet delay, the packet delay variation can be calculated from the packet delay data.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a schematic diagram showing a network configuration of a packet delay data processing system according to an embodiment of the present invention.
The packet delay data processing system includes a packet delay data processing apparatus 1 and a plurality of measurement terminals 2 connected to a network that is a correction target of packet delay data. The plurality of measurement terminals 2 are connected to each other via a network. Further, each measurement terminal 2 and the packet delay data processing device 1 can communicate with each other via, for example, the Internet. In the present embodiment, the packet delay is measured by M paths P _j (j = 1, 2,..., M) between N measurement terminals 2.

  Each measuring terminal 2 generates packet delay data of the received packet, and transmits the generated packet delay data to the packet delay data processing device 1. The packet delay data is data representing the delay time of the packet, and is generated from the difference between the time when the measuring terminal 2 receives the packet and the transmission time embedded in the packet. The packet delay data processing device 1 corrects the packet delay data received from each measurement terminal 2.

FIG. 2 is a block diagram showing a configuration of the packet delay data processing apparatus 1 in the present embodiment.
The packet delay data processing device 1 includes a control unit 11 that controls and controls the packet delay data processing device 1, a storage unit 12 that stores packet delay data received from each measurement terminal 2, and communication with each measurement terminal 2. The communication unit 13 is configured to include a packet delay data correction unit 14 that corrects the packet delay data stored in the storage unit 12.

FIG. 3 is a flowchart showing an example of processing of the packet delay data correction unit 14 in the present embodiment.
First, in step S <b> 1, the packet delay data correction unit 14 acquires packet delay data stored in the storage unit 12.
In the next step S2, the packet delay data correction unit 14 decomposes the acquired packet delay data into a high frequency component and a low frequency component by wavelet transform or Fourier transform. Specifically, by subtracting one of the low frequency range or high frequency range of the frequency component obtained as a result of the Fourier transform, and performing inverse Fourier transform / inverse wavelet transform, the packet delay data of the high frequency component and the low frequency component are respectively obtained. get. At this time, the extracted high-frequency component can be approximated as a packet delay variation.
In the next step S3, the packet delay data correcting unit 14 calculates the average value ave _j (j = 1, 2,..., M) and the variance value disp _j (j = 1, 2, M) of the low frequency components of each path P _j . ..., M) is calculated. Here, when each measuring terminal 2 is time-synchronized, the average value ave _j calculated from sufficiently long packet delay data can be approximated to a fixed component of the packet delay.

  Although packet delay data discontinuity may occur due to a route change or the like, the influence of a temporary route change can be reduced by performing measurement for a sufficiently long time. Alternatively, when measurement is performed for a short time, packet delay data whose path has been changed based on a change in the low frequency component of the packet delay data is excluded from the correction target. Further, correction processing can be performed by grouping on the same route.

In the next step S4, the packet delay data correction unit 14 determines whether or not all the calculated average values ave _j satisfy a preset constraint condition. The constraint condition is that “ave _j > 0” in all paths. Furthermore, it is also possible to specify a range of the average value ave ₁ of a specific path as a constraint condition. For example, when the path P _l performs optical communication and the measuring terminal 2 that is both ends of the path P _l is installed at a distance of 1000 km, “ave _l > 20 ms due to transmission delay characteristics of the optical fiber and the like. Specify additional constraints such as If the constraint condition is satisfied, the process proceeds to step S6. On the other hand, if the constraint condition is not satisfied, the process proceeds to step S5.

In step S5, the packet delay data correction unit 14 optimizes the average value ave _j of the packet delay data by nonlinear programming. Details of processing by this nonlinear programming will be described later.
In the next step S6, the packet delay data correction unit 14 adds the high frequency component to the calculated average value ave _j of the packet delay data, and ends the processing using the corrected packet delay data as the corrected value.

FIG. 4 is a schematic diagram showing an outline of the packet delay average value optimization processing by the nonlinear programming method in the present embodiment.
The packet delay average correction process by the non-linear programming method is executed by the packet delay data correction unit 14 in step S5.
FIG. 4 shows true packet delays in paths (P ₁ , P ₂ , P ₃ ) between a plurality of measurement terminals 2 (node N ₁ , node N ₂ , node N ₃ ) that have time lags, and the respective measurements. A packet delay measurement result (packet delay data) measured by the terminal 2 is shown. True packet delay is the time from when a packet is transmitted from the transmitting terminal until it is received by the receiving terminal. In the example shown in FIG. 4, the true packet delay of the path P ₁ from the node N ₁ to the node N ₃ is 4 ms, but the measurement result is −12 ms. Also, the true packet delay path _{P 2} from the node _{N 2} to the node _{N 1} is 10ms but the measurement result is 9 ms. The true packet delay of the path P ₃ from the node N ₂ to the node N ₃ is 3 ms, but the measurement result is −8 ms. This is because the node N ₁ is +2 ms, the node N ₂ is +1 ms, the node N ₃ is −10 ms, and the time is shifted from the true time. Since the measurement result is smaller than 0 in the path P ₁ and the path P ₃ , the above-described constraint condition “ave _j > 0” is not satisfied. For this reason, it can be seen from the packet delay data that the nodes N ₁ , N ₂ and N ₃ are not time synchronized. Therefore, the packet delay data correction unit 14 performs optimization to bring the packet delay data closer to the true packet delay.

Next, a nonlinear programming problem used for optimization will be described. In general, the nonlinear programming problem is a problem of minimizing the objective function Z under the following constraint conditions. Many solution algorithms have been proposed for nonlinear programming problems. Here, A is an N × M matrix, b and d are N + M dimensional vectors, C is a vector, and D is an N × N symmetric matrix.
Constraints: Ab = C, Ab ≧ C
The objective ^{function: Z = -d T b + (} 1/2) b T Db

First, the constraint conditions in this embodiment will be described. Shift time in each measurement terminal 2 (the time deviates from the true _{time) X k (k = 1,2,} ..., N), the fixed component of the true packet delay of each path _{P j X N + j} If (j = 1, 2,..., M), the following constraint condition 1 is established using the calculated average value ave _j . Here, the transmission terminal of path P _j is N _S , and the reception terminal is N _r . Constraint 1 is that “(true packet delay) − (time difference of transmitting terminal) + (time difference of receiving terminal)” is a value close to “(measured packet delay)”.
(Constraint 1) X _{N + j} −X _s + X _r ≈ave _j

Further, the following constraint condition 2 is satisfied. Constraint 2 is fixed components X _{N + j} of the true packet delay of all paths P _j is to be non-negative.
(Constraint 2) X _{N + j} (j = 1, 2,..., M) ≧ 0

Further, when there is a restriction on the fixed component X _{N + l} of the true packet delay of the specific path P _l, the following restriction condition 3 is added. delay1 and delay2 are preset constants.
(Constraint 3) X _{N + l} ≧ delay ₁ , X _{N + l} ≦ delay 2

Since the constraints 1 to 3 are expressed by linear simultaneous equations / inequality, A is a coefficient matrix of N + M columns, C = [ave ₁ , ave ₂ ,..., Ave _M , 0, 0,..., 0, delay 1,. Then, it can be expressed as Ab = C, Ab ≧ C.

Next, a method for calculating the objective function will be described. First, the packet delay data correction unit 14 calculates the dispersion value disp _j of the low frequency component of the packet delay data of all paths, and calculates the instability I _k of each measurement terminal 2. The instability I _k is calculated by the following equation (1). Expression (1) indicates that the degree of anxiety I _k is defined as an average value of the variance values of paths in which N _k is a packet transmitting terminal or receiving terminal.

Next, the objective function Z is defined by, for example, the following equation (2) using the calculated instability I _k . δ is a preset constant.

  That is, D is the following equation (3), and d is the equation (4).

Next, the packet delay data correction unit 14 minimizes the defined objective function Z under constraint conditions, X _k (k = 1, 2,..., N) and X _{N + j} (j = 1, 2,..., M) is determined by nonlinear programming. The packet delay data correction unit 14 applies, for example, a sequential quadratic programming method as a typical solution. As a result, the higher the measurement terminal 2 uneasiness about I _k time deviation (squared) are largely corrected. As a result, the fixed component X _{N + j} of the true packet delay is calculated. The fixed component X _{N + j} of the packet delay is a value obtained by optimizing the average value ave _j of the packet delay data. The packet delay data correction unit 14 calculates corrected packet delay data by adding a high frequency component to the average value ave _j of the optimized packet delay data.

As described above, according to the present embodiment, the packet delay data is subjected to wavelet transform or Fourier transform to be decomposed into a high frequency component and a low frequency component, and an average value of the decomposed low frequency components is calculated for each path. Since the average value of the low frequency components can be a fixed component of the packet delay, the fixed component of the packet delay can be linearly corrected. Thereby, it is possible to apply the linear correction and perform the correction with high accuracy. Further, since the high frequency component can be a variation in packet delay, the packet delay variation can be calculated from the packet delay data.
Further, it is determined whether or not the constraint condition is satisfied. If not satisfied, the average value of the low frequency components is optimized by nonlinear programming so as to satisfy the constraint condition. Since nonlinear programming is used, it is possible to optimize the packet delay data in which the time lag of each measuring terminal 2 is corrected. As a result, multipoint packet delay data measured between terminals with low time synchronization accuracy can be corrected.

Further, the program for realizing each step shown in FIG. 3 is recorded on a computer-readable recording medium, and the program recorded on the recording medium is read into a computer system and executed, thereby executing packet delay data correction. Processing may be performed. Here, the “computer system” may include an OS and hardware such as peripheral devices.
Further, the “computer system” includes a homepage providing environment (or display environment) if a WWW system is used.
The “computer-readable recording medium” means a flexible disk, a magneto-optical disk, a ROM, a writable nonvolatile memory such as a flash memory, a portable medium such as a CD-ROM, a hard disk built in a computer system, etc. This is a storage device.

Further, the “computer-readable recording medium” means a volatile memory (for example, DRAM (Dynamic DRAM) in a computer system that becomes a server or a client when a program is transmitted through a network such as the Internet or a communication line such as a telephone line. Random Access Memory)), etc., which hold programs for a certain period of time.
The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.
The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

  As described above, the embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to the above, and various design changes and the like can be made without departing from the scope of the present invention. It is possible to

It is the schematic which shows the network structure of the delay data processing system by one Embodiment of this invention. It is a block diagram which shows the structure of the packet delay data processing apparatus in this embodiment. It is the flowchart which showed an example of the process of the packet delay data correction | amendment part in this embodiment. It is the schematic which showed the outline | summary of the average value optimization process of the packet delay by the nonlinear programming in this embodiment.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Packet delay data processing apparatus 2 ... Measuring terminal 11 ... Control part 12 ... Memory | storage part 13 ... Communication part 14 ... Packet delay data correction part

Claims (9)

A packet delay data processing device for correcting packet delay data between a plurality of terminals distributed in a network,
Storage means for storing packet delay data of each path between the terminals;
Decomposition means for decomposing the packet delay data stored in the storage means into a high frequency component and a low frequency component by Fourier transform or wavelet transform;
Calculating means for calculating an average value of the decomposed low frequency components for each path;
Adding means for adding high-frequency components to the average value to obtain corrected packet delay data;
A packet delay data processing apparatus comprising:   2. The packet delay data processing apparatus according to claim 1, further comprising optimization means for optimizing an average value of the low frequency components of all paths so as to satisfy a predetermined constraint condition by nonlinear programming.   3. The packet delay data processing apparatus according to claim 2, wherein the optimization unit uses an objective function defined using instability of each terminal based on a dispersion value of packet delay data of each terminal. . A packet delay data processing method comprising: storing means for storing packet delay data of a path between terminals, and correcting packet delay data between a plurality of terminals distributed in a network,
Decomposing the packet delay data stored in the storage means into a high frequency component and a low frequency component by Fourier transform or wavelet transform;
Calculating an average value of the low frequency components;
Adding a high frequency component to the average value to obtain corrected packet delay data; and
A packet delay data processing method comprising:   5. The packet delay data processing method according to claim 4, further comprising an optimization step of optimizing an average value of the low frequency components of all paths by a non-linear programming method so as to satisfy a predetermined constraint condition.   6. The packet delay data processing method according to claim 5, wherein the optimization step uses an objective function defined using instability of each terminal based on a dispersion value of packet delay data of each terminal. . A program for correcting packet delay data between a plurality of terminals distributed in a network,
A computer comprising storage means for storing packet delay data of the path between the terminals,
Decomposing the packet delay data stored in the storage means into a high frequency component and a low frequency component by Fourier transform or wavelet transform;
Calculating an average value of the low frequency components;
Adding a high frequency component to the average value to obtain corrected packet delay data; and
A program for running   The program according to claim 7, further comprising an optimization step of optimizing an average value of the low frequency components of all paths so as to satisfy a predetermined constraint condition by nonlinear programming.   9. The program according to claim 8, wherein the optimization step uses an objective function defined using the instability of each terminal based on a dispersion value of packet delay data of each terminal.

Images