JP2008017407A - Network quality measuring instrument and method thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a network quality measuring instrument capable of measuring the quality even with a measure of low throughput. <P>SOLUTION: The network quality measuring instrument includes: means 111, 112, 113 for acquiring data from a network; a means 114 for specifying a flow to which the data belong, based on partial fields of the data; means 301b, 302b, 303b for counting a specific event from a transmission order number of the data or from transitions of the transmission order number and an acknowledgement signal number; a means 305b for calculating a detection probability of counted events; and a means 305b for calculating the number of omitted data or a data omission rate that occurs between the measure and a receiving terminal from the counted events and the detection probability. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an apparatus for measuring network quality and a communication quality measuring method used therefor.

  FIG. 1 is a diagram illustrating an application area of the present technology.

  The present invention relates to an apparatus and method for measuring network quality. The quality of the network here refers to the quality of the data input to the measuring device 1a, that is, the quality of the data passing through the branching device 4 and the branching device 5, and the quality of the data is “throughput / goodput”, “End-to-End packet loss”, “transmission terminal-measuring device packet loss”, “measuring device-receiving terminal packet loss”, “measuring device packet loss”.

  In the following, for simplification, an apparatus for measuring network quality using TCP will be described.

  As a technique for measuring the quality of TCP communication, the following method (Yasuhiro Yamazaki, Hideyuki Shimonishi, Tsutomu Murase, “Communication Bottleneck Diagnosis Method”, 2004 Society of Electronics, Information and Communication Engineers Society Conference (BS-3-2), (September 2004) (Non-Patent Document 1) exists.

  FIG. 2 is a block diagram showing the configuration of a measuring apparatus 1a according to the conventional invention.

  The measuring apparatus 1a of the conventional example includes a data receiving unit 111 that is data input from a branching device, a data receiving unit 112, a filtering unit 120 that identifies input data for each same flow, and a current network state. Checked status monitoring unit 130a, ACK number monitoring unit 131a for monitoring the movement of the acknowledgment signal (ACK) number therein, retransmission monitoring unit 132a for monitoring the number of duplicate ACK numbers, and received so far A maximum SN monitoring unit 133a for storing the maximum sequence number (SN), an SN monitoring unit 134a for monitoring SN movement, a quality determination unit 140a for determining a packet loss and a packet loss occurrence point, A packet loss determination unit 141a for determining whether or not a packet in the packet has lost, and, as a result, communication determined to be normal communication The normal communication 142a for storing the packet, the packet loss storage unit 143a for storing the communication amount determined to be packet loss, and the amount of loss between the measuring apparatus 1a and the communication terminal 2 in the packet loss A communication terminal 2-measuring device 1a packet loss storage unit 1431a, a communication terminal 3-measuring device 1 packet loss storage unit 1432a that stores the amount of loss between the measuring device 1a and the communication terminal 3 in the packet loss, Consists of

  In this conventional example, the processing is started by capturing data flowing on the network by the measuring device 1a. Data input from the branch device 4 is received by the data receiving unit 111, and data input from the branch device 5 is received by the data receiving unit 112. After receiving data by the data receiving unit 111 and the data receiving unit 112, the receiving unit passes the data to the filtering unit 120. The filtering unit 120 identifies the flow of the received data based on the transmission / reception IP address, transmission / reception TCP port number, protocol number, and the like, and notifies the state monitoring unit 130a of the result. The state monitoring unit 130a monitors, for each same flow, the SN that is the data order information in the TCP protocol header of the received data and the ACK number that describes the SN number that has been confirmed to be received by the receiving terminal. Information of the unit 131a, the retransmission monitoring unit 132a, the maximum SN monitoring unit 133a, and the SN monitoring unit 134a is extracted. The ACK number monitoring unit 131a checks the ACK number measured so far, the retransmission monitoring unit 132a checks the number of times the currently measured maximum ACK number has been received, and the maximum SN monitoring unit 133a The SN monitoring unit 134a performs a check process of the number of times of measurement with the same number as the measured SN number until the maximum SN check process until a loss occurs. After the processing of the state monitoring unit 130a is completed, the process moves to the quality determination unit 140a. In the processing in the quality determination unit 140a, first, the packet loss determination unit 141a looks at the state of the state monitoring unit 130a to determine whether or not a packet has been lost. Identify. Based on the determination result, if no packet loss has occurred, the communication amount is added to the normal communication 142a. When a packet loss has occurred, the loss amount is recorded in the packet loss storage unit 143a. Further, based on the result of the occurrence location of the packet loss, the amount of packet loss is recorded in the communication terminal 2 -measurement device 1 packet loss storage unit 1431 a or the measurement device 1 -communication terminal 3 packet loss storage unit 1432 a. .

  For the sake of simplicity, this conventional example has been described with a device that measures the quality of TCP communication. However, the sequence of data strings is described in the transmission data, and there is a mechanism for retransmission for data loss. Is a common technology. Therefore, it includes general protocols in which retransmission mechanisms such as HSTCP, SCTP, and DCCP exist.

As an application area, not only the state of FIG. 1 that does not affect the network to be measured and traffic, but also a format that can acquire data is inserted in the middle between communication terminals, and the network to be measured and traffic are affected. The form shown in FIG. 3 is also possible. The data relay terminal in FIG. 3 is an Ethernet switch that performs data transfer at Layer 2, a router that performs data transfer at Layer 3, a gateway that performs transfer at Layer 4 or higher, or the like. This refers to a terminal that has been transferred with a changed protocol or that has a load balance function or bandwidth control function.
JP 2006-067217 A International Publication WO / 2006/043624 Yasuhiro Yamazaki, Hideyuki Shimonishi, Tsutomu Murase, “Communication Bottleneck Diagnosis Method”, Society Conference of the Institute of Electronics, Information and Communication Engineers (BS-3-2), September 2004 Yoshihiro Tsurugazawa, Nobuo Kawamura, Noriko Yoshimura, “TCP Transfer Quality Degradation Detection and Degradation Interval Method”, 2005 IEICE General Conference (B-11-14), March 2005 Yoshihiro Tsurugazawa, Toshiaki Tsuchiya, Nobuo Kawamura, “Degradation detection and segmentation of IP packet transfer quality”, 2005 IEICE Society Conference (B-11-8), September 2005 Yasuhiro Yamazaki, Hideyuki Shimonishi, Tsutomu Murase, “Proposal of Packet Loss Rate Estimation Method by Sampling Measurement”, IEICE Technical Report (NS2004-159, TM2004-62), November 2004 Yasuhiro Yamazaki, Hideyuki Shimonishi, Tsutomu Murase, “TCP packet loss monitoring technology in high-speed networks”, IEICE Technical Report (TM2005-11), May 2005

  The first problem is that the conventional method requires high processing capability in order to measure the quality of data.

  The reason is that the conventional method needs to process all the packets of the flow to be measured. For this reason, in a high-speed network, the number of packets to be calculated becomes enormous, and it is difficult to capture all the packets. Even when all the packets can be captured, a high processing capacity is required.

  The second problem is that the conventional method cannot measure the quality correctly in a situation where all the packets of the flow to be measured cannot be acquired.

  The reason is that in the conventional method, packets such as “End-to-End packet loss”, “Transmission terminal-measuring device packet loss”, “Measuring device-receiving terminal packet loss”, and “Measuring device packet loss” are used. This is because the loss count is determined based on the continuity of sequence numbers, and if any sequence number is missing, it is classified as one of the packet losses. For this reason, in a situation where all the packets of the corresponding flow cannot be acquired due to dropping or the like, even if no packet loss has occurred, the packet is classified as one of the packet losses, and the determination is erroneous. For this reason, quality cannot be measured correctly.

  The object of the present invention has been devised in view of the above problems, and in a measuring instrument, “throughput / goodput”, “end-to-end packet loss”, “transmission terminal-measuring device packet loss”. To provide a communication quality measurement system capable of measuring network quality such as “packet loss between measuring device and receiving terminal” and “packet loss between measuring devices” based on some packets belonging to the flow. It is in.

  Accordingly, it is possible to provide a network quality measuring apparatus and method capable of measuring quality even with a measuring instrument having low processing capability because quality measurement is possible without requiring all packets.

  It is another object of the present invention to provide a network quality measuring apparatus and method capable of measuring quality even when a packet is lost because quality measurement is possible from some packets belonging to a flow.

  In the measuring apparatus 1 according to the present invention, the sampling unit performs quality measurement on the thinned packets. As a result, it is not necessary to perform measurement processing on all packets, and the measuring instrument does not need high calculation capability.

  In the measuring apparatus 1 according to the present invention, in order to be able to measure the end-to-end communication quality of the network by sampling measurement, the ACK order determining unit 201b, the duplicate ACK determining unit 202b, and the end-to-end information storage unit The throughput and end-to-end packet loss are calculated by 205b, the throughput measuring unit 204b and the end-to-end loss determining unit 203b. As a result, correct end-to-end communication quality can be calculated even when sampling measurement is performed.

  In the measuring apparatus 1 according to the present invention, in order to be able to measure the end-to-end communication quality of the network by sampling measurement, the SN order determining unit 201c, the SN degradation determining unit 202c, and the end-to-end information storage unit The throughput and the end-to-end packet loss are calculated by 205c, the throughput measuring unit 204c and the end-to-end loss determining unit 203c. As a result, correct end-to-end communication quality can be calculated even when sampling measurement is performed.

  In the measurement apparatus 1 according to the present invention, the SN information management unit 301b is used to measure the network communication quality “packet loss between the transmission terminal and the measurement apparatus” and “packet loss between the measurement apparatus and the reception terminal” by sampling measurement. And a duplicate ACK determination unit 302b, a passing SN storage unit 303b, a target event number recording unit 304b, a measurement device-receiving terminal loss determination unit 305b, and a transmission terminal-measurement device loss determination unit 306b. It is carried out. As a result, even when sampling measurement is performed, it is possible to calculate the correct “transmission terminal-measuring device packet loss” and “measuring device-receiving terminal packet loss”.

  In the measuring apparatus 1 according to the present invention, the SN information management unit 301c is used to measure the network communication quality “packet loss between transmitting terminal and measuring apparatus” and “packet loss between measuring apparatus and receiving terminal” by sampling measurement. Calculation from the re-SN confirmation unit 302c, the passing SN storage unit 303b, the target event number recording unit 304c, the measurement device-receiving terminal loss determination unit 305c, and the transmission terminal-measurement device loss determination unit 306b. It is carried out. As a result, even when sampling measurement is performed, it is possible to calculate the correct “transmission terminal-measuring device packet loss” and “measuring device-receiving terminal packet loss”.

  In the measuring apparatus 1 according to the present invention, in order to be able to measure the network communication quality “packet loss between transmitting terminal and measuring apparatus” and “packet loss between measuring apparatus and receiving terminal” by sampling measurement, the detection probability calculating unit 308d. And the adoption loss determination unit 307d, the packet loss direction identification unit 300b, and the packet loss direction identification unit 300c. As a result, even when sampling measurement is performed, it is possible to measure network communication quality with higher accuracy. Here, the packet loss direction is a direction input to the measurement device or a direction output from the measurement device.

  In the quality measurement system according to the present invention, in order to enable measurement of network communication quality “inter-measuring packet loss” by sampling measurement, a plurality of measurement devices such as a measurement device 1111, a measurement device 1112, and a measurement device 111N, The determination is made from the measurement server 2000 that calculates the result. As a result, even when performing sampling measurement, it is possible to measure “inter-measuring packet loss” with higher accuracy.

  According to the present invention, the communication quality measurement system does not require high computing capacity.

  This is because according to the present invention, it is possible to calculate the communication quality index only by acquiring a part of the packet sequence, so that the number of packets to be calculated in the quality measurement system can be greatly reduced. is there.

  The best mode for carrying out the present invention will be described below in detail with reference to the drawings.

  FIG. 4 is a block diagram showing the configuration of the first embodiment of the measuring apparatus 1b according to the present invention.

The measuring device 1b according to the first embodiment has a data reception unit 111 and a data reception unit 112, which are data inputs from the branching device, and packets received by the data reception units 111 and 112 with a probability of 1 in N packets. A data preprocessing unit 100a including a sampling processing unit 113 that acquires information, a flow identification unit 114 that performs flow classification using information in the acquired packet, and a flow information storage unit 115 that stores the result. When,
An ACK order determination unit 201b that extracts information from the flow information storage unit 115 and determines the magnitude of an acknowledgment signal (ACK) number, and a duplicate ACK determination that extracts information from the flow information storage unit 115 and determines the continuity of the ACK number Unit 202b, an end-to-end information storage unit 205b for storing the results, a throughput determination unit 204b for calculating the throughput from the storage information 205b, and determining the number of end-to-end losses / rate from the storage information An end-to-end quality estimation unit 200b including an end-to-end loss determination unit 203b;
An SN information management unit 301b that extracts a sequence number (SN) of a passing packet from the flow information storage unit 115, a passing SN storage unit 303b that stores the result, and an ACK number that extracts information from the flow information storage unit 115 A duplicate ACK determination unit 302b that checks the relationship between the continuity and SN, a target event storage unit 304b that stores the number of times that a check target event has occurred, and a packet that occurs between the measuring instrument and the receiving terminal based on that information Measuring device-receiving terminal loss determination unit 305b for calculating the number of losses / rate, and determining the number / rate of transmission terminal-measuring device loss from end-to-end loss information and measuring device-receiving terminal loss information. A packet loss direction specifying unit 300b including a transmission terminal-measuring device loss determination unit 306b,
A quality result storage unit 400 that stores the result of the end-to-end quality estimation unit 200b and the result of the packet loss direction identification unit 300b is configured.

  In the present embodiment, first, a packet flowing on the network is captured by the measuring device 1b, and the process is started by entering the data preprocessing unit 100a. Data input from the branch device 4 is received by the data receiving unit 111, and data input from the branch device 5 is received by the data receiving unit 112. After receiving data by the data receiving unit 111 and the data receiving unit 112, the receiving unit passes the data to the sampling processing unit 113. The sampling processing unit 113 acquires packet information with a probability of one in N. As a sampling processing method at this time, a random number is generated and compared with a packet acquisition probability (1 / N), and a random sampling is performed to determine whether or not the packet is discarded. One counter is used for each packet, each protocol, and each flow. A method of comparing the sampling value with a part of the packet header information such as a sequence number or an acknowledgment signal number, etc. . After the acquisition packet is determined by the sampling processing unit 113, the flow identification unit 114 uses a part of the information of the acquisition packet, one or more information of IP address, port number, protocol ID, VLAN information, and MPLS information. The packets are classified (flow identification), and the result is stored in the flow information storage unit 115.

  Here, the data preprocessing unit 100a performs the processing of the flow identification unit 114 after performing the sampling processing unit 113. After the flow identification processing unit 114 identifies the flow, the sampling processing unit 113 determines an acquisition packet. It doesn't matter. Further, the result of sampling by the data preprocessing unit will be used in the end-to-end quality estimation unit 200b and the packet loss direction specification unit 300b in the future, but the packet has the sampling pattern for the end-to-end quality estimation unit 200b. The packet loss direction specifying unit 300b can also determine whether to discard the packet using another sampling pattern and store it separately in the flow information storage unit 115.

The end-to-end quality estimation unit 200b performs end-to-end communication quality analysis at a specified time range or at certain intervals. The ACK order determination unit 201b sequentially takes out the ACK side packets of the quality analysis target (packets corresponding to the specified flow in the specified period) from the flow information storage unit 115, and determines the size of the ACK number. The ACK numbers are sequentially compared to find the minimum ACK number and the maximum ACK number, and the result is recorded in the end-to-end information storage unit 205b. In addition, the duplicate ACK determination unit 202b designates the ACK duplication degree (k) in advance and checks whether or not the designated ACK duplication degree (k) is exceeded each time the duplicate ACK phenomenon occurs. The total number of duplicate ACKs in the -to-End information storage unit 205b is counted up (see FIG. 5, the degree of duplication is 5 in the example in the figure, and the count is incremented if k = 2. If k = 6) Does not count up). When the processing in the ACK order determination unit 201b and the processing in the duplicate ACK determination unit 202b are completed for all the quality analysis target packets, the throughput determination unit 204b and the end-to-end loss determination unit 203b are processed. Throughput determination unit 204b can determine the communication volume and throughput during the period from the difference information of the ACK number as shown in equation (1). Further, the end-to-end packet loss is calculated by calculating a percentage of the duplicate ACK when the duplicate ACK that can be observed by sampling is not sampled, as shown in the equation (2), and the packet when not sampling from the duplicate ACK total number information. The loss amount x _i is calculated. The expression (2) is described in detail in Non-Patent Documents 4 and 5. The calculated result of the throughput determination unit 204b and the result of the end-to-end loss determination unit 203b are recorded in the quality result storage unit 400.

  Data traffic = Maximum ACK number-Minimum ACK number (1)

Equation (2) obtains a solution by iterative calculation. Here, s is a sampling probability, k is a threshold value of ACK duplication degree (counting the number of duplicate ACK phenomena exceeding this duplication degree), Th is data traffic, A _k is the total number of duplicate ACKs, and b is a delay ACK of TCP. It is a parameter.

  Here, the end-to-end packet loss calculation only needs to know the ratio of the total number of duplicate ACKs that can be observed under sampling to the total number of duplicate ACKs that should be observable when sampling is not performed (detection probability 1). Therefore, it is not always necessary to calculate using the equation (2), the network is checked in advance, and the detection probability table 1 (packet loss rate, detection probability 1) at each sampling probability is checked to create a database. (DB) may be created, and the following (3) and (4) calculations may be repeated several times to calculate an end-to-end packet loss.

  Recognize Pi as packet loss, and obtain detection probability Qi at the time of corresponding packet loss in detection probability table 1. Calculate (4).

  The value obtained by the calculation of (4) is substituted into (3) again, and this calculation is repeated several times.

(2) and (3), the initial value of Xi in equation (4), it is conceivable to use an A value and packet loss count immediately before the _k as an example, also using an appropriate value solutions Can be obtained.

  The packet loss direction identification unit 300b also performs packet loss direction identification processing at a specified time range or at certain intervals. Here, the start time and the end time of the quality analysis do not necessarily have to completely coincide with the end-to-end quality estimation unit 200b. The SN information management unit 301b extracts the data-side packet of the corresponding flow from the flow information storage unit 115, and stores the existing sequence number in the passing SN storage unit 303b. When the sequence number decreases, registration is not performed in the passing SN storage unit 303b. The duplicate ACK determination unit 302b also extracts the ACK side packet of the corresponding flow from the flow information storage unit 115, and checks whether a duplicate ACK phenomenon with a specified redundancy (k) or more has occurred, as with the duplicate ACK determination unit 202b. . When the duplicate phenomenon has occurred, the passing SN storage unit 303b is referred to and it is checked whether or not the sequence number having the number that is retransmitted by the duplicate ACK exists in the table (see FIG. 6). ). When it exists in the table, it recognizes that the target event exists, counts up the number of target events, and stores it in the target event storage unit 304b. The duplicate ACK determination unit deletes the sequence number for which the check has been completed (smaller than the ACK number for which the observation has been confirmed) from the passing SN storage unit 303b. When the processing in the SN information management unit 301b and the processing in the duplicate ACK determination unit 302b are completed for all packets subject to quality analysis, the loss determination unit 305b between the measuring instrument and the receiving terminal and the loss determination unit 306b between the transmission terminal and the measuring device I do. In the measuring instrument-receiving terminal loss determination unit 305b, the end-to-end packet loss value (P) of the corresponding flow is obtained from the quality result storage unit 400, and between the measuring instrument-receiving terminal is calculated by the equation (5). Calculate the loss count x0.

  Here, Bk is the number of target events recorded in the counted target event storage unit 304b. In addition, the value of the loss between the measuring instrument and the receiving terminal here is obtained if the total number of target events that can be observed at the time of sampling is known as a ratio of the total number of target events that can be observed when the sampling is not performed (detection probability 2). So, check the network in advance, check the detection probability table 2 (packet loss rate, detection probability 2) at each sampling probability, create a database (DB), and compare the end-to-end packet loss rate with quality After obtaining from the result storage unit 400, the detection probability table 2 may be examined to obtain the detection probability 2, and the packet loss between the measuring instrument and the receiving terminal may be calculated by calculating the equation (6).

  After calculating the packet loss between the measuring instrument and the receiving terminal by the equation (6), the packet loss between the transmitting terminal and the measuring instrument can be calculated by subtracting the value from the end-to-end packet loss.

  Next, the flow of processing in the measuring device 1b will be described with reference to FIGS.

  The overall flow of the measuring device 1b is as follows. After the processing of the data preprocessing unit 100a (FIG. 8), the processing of the end-to-end quality estimation unit 200b (based on the information classified / stored there ( 9) and the processing of the packet loss direction specifying unit 300b (FIG. 10) are started.

  Here, FIG. 8 shows an outline of a processing flow in the data preprocessing unit 100a.

  The data preprocessing unit 100 a starts processing when data is input from the branching device 4 or the branching device 5 and arrives at the data receiving unit 111 and the data receiving unit 112. This process is process A-1. After this process is completed, the process moves to process A-2.

  Process A-2 is a packet sampling (thinning-out) process. The sampling processing unit 113 acquires packet information with a probability of one in N. As a sampling processing method at this time, a random number is generated and compared with a packet acquisition probability (1 / N), and a random sampling is performed to determine whether or not the packet is discarded. One counter is used for each packet, each protocol, and each flow. A method of comparing the sampling value with a part of the packet header information such as a sequence number or an acknowledgment signal number, etc. . After the sampling process ends, the process proceeds to process A-3.

  In the processing A-3, after the acquisition packet is determined by the sampling processing unit 113, the flow identification unit 114 selects any one of information of the acquisition packet, IP address, port number, protocol ID, VLAN information, and MPLS information. The packet is classified (flow identification) using the above information, and the result is stored in the flow information storage unit 115.

  FIG. 9 shows an outline of a processing flow in the end-to-end quality estimation unit 200b.

  Processing is started by fetching the ACK side packet of the quality measurement target flow from the flow information storage unit 115 into the End-to-End quality estimation unit 200b. This process is process B-1. After this process is completed, the process moves to process B-2.

  In the process B-2, the ACK order determination unit 201b determines the ACK number of the packet captured so far during the quality measurement target period. Information stored as a minimum ACK number in the past is extracted from the information storage unit 205b and compared with the currently acquired ACK number using equation (7).

Current ACK number <Information previously stored as the minimum ACK number (7)
If the expression (7) is true, the process proceeds to process B-3. If it is false, the process proceeds to process B-4.

  In process B-3, the currently acquired ACK number is recognized as the minimum ACK number, stored in the information storage unit 205b, and the process proceeds to process B-4.

  In the process B-4, the ACK order determination unit 201b determines the ACK number of the packet captured so far during the quality measurement target period. Information stored as the maximum ACK number in the past is extracted from the information storage unit 205b, and the currently acquired ACK number is compared with equation (8).

Current ACK number> Information previously stored as the maximum ACK number (8)
If the expression (8) is true, the process moves to process B-5. If it is false, the process proceeds to process B-6.

  In process B-5, the currently acquired ACK number is recognized as the maximum ACK number, stored in the information storage unit 205b, and the process proceeds to process B-6.

  In the process B-6, the ACK number stored in the past is compared with the ACK number captured this time by the duplicate ACK determination unit 202b. For the determination of duplicate ACK, the ACK number obtained immediately before is compared with the ACK number acquired this time, and if the number is the same, the degree of duplication is counted up (+1). In the system, first, the degree of duplication recognized as a duplicate ACK is set as k, and k is compared with the degree of duplication. When “k = redundancy”, the process moves to process B-7, and when “k ≠ duplication degree”, the process moves to process B-8. If the ACK number stored in the past does not match the ACK number fetched this time, “duplication degree = 0” is set.

  In the process B-7, the total number of duplicate ACKs stored in the information storage unit 205b is counted up (+1). The total number of duplicate ACKs is cleared (= 0) at the start of the observation period. After updating the total number of duplicate ACKs, the process proceeds to process B-8.

  In the process B-8, it is confirmed whether all packets in the observation period target period have been read by the ACK order determination unit 201b and the duplicate ACK determination unit 202b. If there is a packet that has not been determined yet, the next packet is read, and the update process of the maximum ACK information, the minimum ACK information, and the total number of duplicate ACKs is returned to process B-1. If all packets have been determined, the process moves to process B-9 to calculate quality information during that period.

  In the process B-9, the throughput determination unit 204b performs a throughput calculation. The calculation is obtained by equation (1) based on the maximum ACK number and the minimum ACK number. After calculating the throughput, the process proceeds to process B-10.

  In the process B-10, the end-to-end packet loss is calculated by the end-to-end loss determination unit 203b. The calculation is by dividing the observed total number of duplicate ACKs by detection probability # 1 (ratio of the total number of duplicate ACKs that could be observed under sampling to the total number of duplicate ACKs that should be observable if not sampled), Can be calculated. This detection probability # 1 can be calculated by assuming a statistical model as shown in equation (2), or by creating a database by examining past network conditions as shown in equations (3) and (4). It may be in the form of calculating in advance. With this process, the process of the end-to-end quality estimation unit 200b is terminated.

  FIG. 10 shows an outline of a processing flow in the packet loss direction specifying unit 300b.

  Processing is started by fetching both the Data side and ACK side packets of the quality measurement target flow from the flow information storage unit 115 into the packet loss direction specifying unit 300b. This process is process C-1. After this process is completed, the process moves to process C-2.

  In process C-2, the SN number management unit 301b manages sequence numbers. If the acquired packet is the ACK side information of the corresponding flow, the process moves to process C-4 without performing any process. If the packet is on the data side of the flow, the process moves to process C-3.

  In the process C-3, the sequence number of the packet acquired by the SN information management unit 301b is recorded in the passing SN storage unit 303b, and the process proceeds to the process C-4.

  In process C-4, the process of the duplicate ACK determination unit 302 is performed. The duplicate ACK determination condition is the same as that for the duplicate ACK determination unit 202b. The designated duplicate k is compared with the actually duplicated value, and when “k = duplicate”, the duplicate ACK is recognized. If a duplicate ACK is recognized, the process proceeds to process C-5. If a duplicate ACK is not recognized, the process proceeds to process C-7. Here, the value of the duplication degree threshold (k) of the duplicate ACK determination unit 202b and the value of the duplication degree threshold (k) of the duplicate ACK determination unit 302b do not necessarily match.

  In the process C-5, the duplicate ACK determination unit 302b refers to the sequence number stored in the passing SN storage unit 303b, and the data whose retransmission is promoted by the duplicate ACK (data having the sequence number of ACK number + 1) Check if you have passed in the past. If it has not passed in the past, it moves to process C-7, and if it has passed in the past, it moves to process C-6.

  In the process C-6, the duplicate ACK determination unit 302b counts up (+1) the target event number Bk stored in the target event number storage unit 304b. This target event number Bk is cleared (= 0) at the start of the observation period. After the count-up process ends, the process moves to process C-7.

  In the process C-7, it is confirmed whether or not all the packets in the observation period target period have been read by the SN information management unit 301b and the duplicate ACK determination unit 302b. If there is a packet that has not been determined yet, the next packet is read, and the process returns to process C-1 to check the passage of the SN number and update the target event count. If all packets have been determined, the process proceeds to process C-8 to calculate quality information during that period.

  In process C-8, the measuring instrument-receiving terminal loss determination unit 305b calculates the measuring instrument-receiving terminal packet loss. The number of packet losses between the measuring instrument and the receiving terminal is the detection probability # 2 (the total number of target events that can be observed when sampling is not performed. It can be calculated by dividing by (ratio). Detection probability # 2 may be calculated by assuming a statistical model as shown in equation (5), or by creating a database by examining past network conditions as shown in equation (6). The calculation format may be used. After this process is completed, the process moves to process C-9.

  In the process C-9, the transmission terminal-measuring device loss determination unit 306b calculates the expression (9) to obtain the number of transmission terminal-measuring device packet losses. With this process, the process of the packet loss direction determination unit 300b is terminated.

(Packet loss between sending terminal and measuring instrument)
= (End-to-End packet loss)
-(Packet loss between measuring instrument and receiving terminal) (9)
The above is the processing content of the measuring device 1b in the first embodiment according to the present invention.

  In the conventional technology, when counting the amount of packet loss that occurs between the measuring instrument and the receiving terminal and the packet loss that occurs between the transmitting terminal and the measuring instrument, all packets are observed and packet loss occurs (duplication Every time an ACK occurs), check whether the lost packet has passed in the past, and whether the packet loss occurred between the measuring instrument and the receiving terminal or whether it occurred between the transmitting terminal and the measuring instrument Was judged. For this reason, when this calculation method is applied to the sampled packet group, all duplicate ACKs cannot be observed, and therefore the number of packet losses generated in End-to-End is estimated to be considerably lower than the original value. It will be. Further, when a duplicate ACK occurs, the sequence number passed in the past is confirmed, and if a packet for which retransmission is promoted is lost, the number of phenomena is regarded as a generated packet loss between the measuring instrument and the receiving terminal. This is because it does not consider whether the packet that is being retransmitted really did not pass due to packet loss or could not be observed by sampling, so the amount of packet loss between the measuring instrument and the receiving terminal and the transmitting terminal-measuring instrument The inter-packet loss is classified completely different from the original.

  On the other hand, in this embodiment, as an end-to-end packet loss measurement method, a statistical method is used to predict the original ACK duplication number from the number of duplicate ACK phenomena observed by sampling measurement, and packet loss is determined. It is recorded. For this reason, even if all duplicate ACKs cannot be detected, the original number of duplicate ACKs can be predicted. Similarly, the original amount when all packets are observed using a statistical method based on the value of the sequence number that has been urged to be retransmitted in the past based on the value observed by sampling. The amount of packet loss generated between the measuring instrument and the receiving terminal is predicted. For this reason, by sampling, even when not all sequence numbers are recorded, it is possible to predict the amount of sequence numbers that have been urged to be retransmitted in the past. Similarly, since the value of the end-to-end packet loss and the value of the packet loss between the measuring instrument and the receiving terminal can be predicted at the time of sampling, the value of the packet loss between the transmitting terminal and the measuring instrument is also predicted. Can do.

  By using these sampling measurement methods, communication quality, “throughput / goodput”, “end-to-end packet loss”, “transmitting terminal—measurement” even in a situation where not all packets of the flow to be measured can be acquired. It is possible to correctly measure “inter-device packet loss” and “measuring device-receiving terminal packet loss”. In addition, since it is not necessary to perform processing on all packets in order to measure the quality of the flow, the measuring instrument does not need high calculation capability.

  In the present embodiment, for the sake of simplification, the description has been given with the apparatus that measures the quality of TCP communication. However, the order of the data sequence is described in the transmission data, and there is a retransmission mechanism for data loss. Is a common technology. Therefore, it includes general protocols having a retransmission mechanism such as BIC-TCP, FastTCP, HSTCP, SCTP, and DCCP.

  As an application area, not only the state of FIG. 1 that does not affect the network to be measured and traffic, but also a format that can acquire data is inserted in the middle between communication terminals, and the network to be measured and traffic are affected. The form shown in FIG. 3 is also possible. The data relay terminal in FIG. 3 is an Ethernet switch that performs data transfer at Layer 2, a router that performs data transfer at Layer 3, a gateway that performs transfer at Layer 4 or higher, or the like. This refers to a terminal that has been transferred with a changed protocol or that has a load balance function or bandwidth control function.

  FIG. 11 is a block diagram showing the configuration of the second embodiment of the measuring apparatus 1c according to the present invention.

The measuring device 1c according to the second embodiment has a probability of one out of every N packets received by the data receiving unit 111, the data receiving unit 112, and the data receiving units 111 and 112, which are data inputs from the branching device. A data preprocessing unit 100a including a sampling processing unit 113 that acquires information, a flow identification unit 114 that performs flow classification using information in the acquired packet, and a flow information storage unit 115 that stores the result. When,
The SN order determination unit 201c that extracts information from the flow information storage unit 115 and determines the size of the sequence number (SN), and the information is extracted from the flow information storage unit 115 to determine whether the SN number has become smaller than before. SN-decision determining section 202c that performs, end-to-end information storage section 205c that stores the results thereof, throughput determination section 204c that calculates throughput from storage information 205c, and the number of end-to-end loss / An End-to-End quality estimation unit 200c configured by an End-to-End loss determination unit 203c that determines a rate;
An SN information management unit 301b that extracts a sequence number (SN) of a passing packet from the flow information storage unit 115, a passing SN storage unit 303b that stores the result, and an ACK number that extracts information from the flow information storage unit 115 A duplicate ACK determination unit 302b that checks the relationship between the continuity and SN, a target event storage unit 304b that stores the number of times that a check target event has occurred, and a packet that occurs between the measuring instrument and the receiving terminal based on that information Measuring device-receiving terminal loss determination unit 305b for calculating the number of losses / rate, and determining the number / rate of transmission terminal-measuring device loss from end-to-end loss information and measuring device-receiving terminal loss information. A packet loss direction specifying unit 300b including a transmission terminal-measuring device loss determination unit 306b,
A quality result storage unit 400 that stores the result of the end-to-end quality estimation unit 200b and the result of the packet loss direction identification unit 300b is configured.

  Here, the data preprocessing unit 100a, the packet loss direction specifying unit 300b, and the quality result storage unit 400 perform the same configuration / processing as in the first embodiment, and thus the description thereof is omitted.

  The end-to-end quality estimation unit 200c according to the present embodiment performs end-to-end communication quality analysis at a specified time range or every certain interval. The SN order determination unit 201c sequentially takes out the Data side packets of the quality analysis target (packets corresponding to the specified flow in the specified period) from the flow information storage unit 115 and determines the size of the SN number. The SN numbers are sequentially compared to find the minimum SN number and the maximum SN number, and the result is recorded in the end-to-end information storage unit 205c. In addition, in the SN decrease determination unit 202c, the degree of decrease in SN (k) is designated in advance, and when the SN acquired this time is lower than the previous number, the degree of decrease in which the current acquired SN is the largest in the past is checked. (Refer to FIG. 12, in the example in the figure, the degree of decrease is 5, and if k = 2, it counts up. If K = 6, it does not count up). When the processing in the SN order determination unit 201c and the processing in the SN decrease determination unit 202c are finished for all the packets subject to quality analysis, the throughput determination unit 204c and the end-to-end loss determination unit 203c are processed. Throughput determination unit 204c can determine the communication volume and throughput during the period from the difference information of the SN numbers as shown in equation (10). Further, the end-to-end packet loss can be calculated from the SN decrease total number by calculating a percentage of the SN number decrease decrease when the SN number decrease phenomenon that can be observed by sampling is not sampled as shown in the equation (11). The amount of packet loss when not sampled is calculated. The calculated result of the throughput determination unit 204 c and the result of the end-to-end loss determination unit 203 c are recorded in the quality result storage unit 400.

  Data traffic = Maximum SN number-Minimum SN number (10)

Equation (11) obtains a solution by iterative calculation. Here, s is a sampling probability, k is a threshold for SN degradation (counting the number of SN degradation phenomena exceeding this degree of degradation), Th is the amount of data communication, C _k is the total SN degradation, and b is a TCP delay ACK parameter. It is.

  Here, in the end-to-end packet loss calculation, the ratio (detection probability 3) of the total number of SN lowering phenomena that can be observed under sampling and the total number of SN lowering phenomena that can be observed without sampling is divided. Therefore, it is not always necessary to calculate using the equation (11). The network is examined in advance, and the detection probability table 3 (packet loss rate, detection probability 3) at each sampling probability is checked. The database (DB) may be created, and the following (12) and (13) calculations may be repeated several times to calculate the end-to-end packet loss.

  Recognize Pi as packet loss, and obtain detection probability Qi at the time of corresponding packet loss in detection probability table 3. (13) is calculated.

  The value obtained by the calculation in (13) is substituted into (12) again, and this calculation is repeated several times.

  As an example, the initial value of Xi in equations (11), (12), and (13) may be the value of Ck or the number of previous packet losses, but the solution can be obtained even if an appropriate value is used. Obtainable.

  Next, the flow of processing in the measuring device 1c will be described.

  The overall flow of the measuring device 1c is as follows. After the processing of the data preprocessing unit 100a (FIG. 8), the end-to-end quality estimation unit 200c (FIG. 14) is performed based on the information classified / stored there. ) And the processing of the packet loss direction specifying unit 300b (FIG. 10) is started.

  Here, the processes of the data pre-processing unit 100a (FIG. 9) and the packet loss direction specifying unit 300b (FIG. 10) perform the same configuration / processing as in the first embodiment, and thus description thereof is omitted.

  Here, FIG. 14 shows an outline of a processing flow in the end-to-end quality estimation unit 200b.

  The processing is started by fetching the Data-side packet of the quality measurement target flow from the flow information storage unit 115 into the End-to-End quality estimation unit 200c. This process is process D-1. After this process is completed, the process moves to process D-2.

  In the process D-2, the SN order determination unit 201b determines the sequence number of the packet captured so far during the quality measurement target period. Information stored as the minimum SN number in the past is extracted from the information storage unit 205c, and the currently acquired SN number is compared with equation (14).

Currently imported SN number <Information previously stored as the minimum SN number (14)
If the expression (14) is true, the process proceeds to process D-3. If it is false, the process proceeds to process D-4.

  In process D-3, the currently acquired SN number is recognized as the minimum SN number, stored in the information storage unit 205b, and the process proceeds to process D-4.

  In the process D-4, the SN order determination unit 201c determines the sequence number of the packet captured so far during the quality measurement target period. The information stored as the maximum SN number in the past is extracted from the information storage unit 205c, and the currently acquired SN number is compared with the equation (15).

SN number currently imported> Information previously stored as the maximum SN number (15)
If the expression (15) is true, the process proceeds to process D-5. In the case of false, it moves to process D-6.

  In the process D-5, the currently acquired SN number is recognized as the maximum SN number, stored in the information storage unit 205c, and the process proceeds to the process D-6.

  In the process D-6, the SN number stored in the past and the SN number captured this time are compared by the SN decrease determination unit 202c. The determination of SN decrease is made by comparing the SN number obtained immediately before and the SN number acquired this time, and when the SN number obtained immediately before is larger, what is the largest SN obtained this time? Check the degree of decline. The threshold value k of the SN decrease degree is set in advance and the decrease degree of the current acquired SN is compared. If the decrease degree of the current acquisition is equal to or greater than the threshold value, it is recognized that the SN decrease phenomenon has occurred. When the SN decrease phenomenon occurs, the process proceeds to process D-7. When the SN decrease phenomenon does not occur, the process proceeds to process D-8.

  In the process D-7, the SN decrease total number stored in the information storage unit 205c is counted up (+1). This SN decrease total is cleared (= 0) at the start of the observation period. After the update process of the total number of SN decreases, the process proceeds to process D-8.

  In the process D-8, it is confirmed whether or not all the packets in the observation period target period have been read by the SN order determination unit 201c and the SN decrease determination unit 202c. If there is a packet that has not been determined yet, the next packet is read, and the process returns to process D-1 in order to update the maximum SN information, the minimum SN information, and the total SN decrease. If all packets have been determined, the process moves to process D-9 to calculate quality information during that period.

  In process D-9, the throughput determining unit 204c performs throughput calculation. The calculation is obtained by equation (10) based on the maximum SN number and the minimum SN number. After calculating the throughput, the process proceeds to process D-10.

  In process D-10, the end-to-end packet loss is calculated by the end-to-end loss determination unit 203c. The calculation is to divide the total number of observed SN degradation by detection probability # 3 (the ratio of the total number of SN degradation phenomena that can be observed under sampling to the total number of SN degradation phenomena that should be observable without sampling). Can be calculated. This detection probability # 3 can be calculated by assuming a statistical model as shown in equation (11), or a database is created by examining past network conditions as shown in equations (12) and (13). However, it may be in the form of calculation. With this process, the process of the end-to-end quality estimation unit 200c ends.

  The above is the processing content of the measuring device 1c in the second embodiment of the present invention.

  In the conventional technology, when counting the amount of packet loss that occurs between the measuring instrument and the receiving terminal and the packet loss that occurs between the transmitting terminal and the measuring instrument, all packets are observed and packet loss occurs (duplication Every time an ACK occurs), check whether the lost packet has passed in the past, and whether the packet loss occurred between the measuring instrument and the receiving terminal or whether it occurred between the transmitting terminal and the measuring instrument Was judged. For this reason, when this calculation method is applied to the sampled packet group, all duplicate ACKs cannot be observed, and therefore the number of packet losses generated in End-to-End is estimated to be considerably lower than the original value. It will be. Further, when a duplicate ACK occurs, the sequence number passed in the past is confirmed, and if a packet for which retransmission is promoted is lost, the number of phenomena is regarded as a generated packet loss between the measuring instrument and the receiving terminal. This is because it does not consider whether the packet that is being retransmitted really did not pass due to packet loss or could not be observed by sampling, so the amount of packet loss between the measuring instrument and the receiving terminal and the transmitting terminal-measuring instrument The inter-packet loss is classified completely different from the original.

  On the other hand, in this embodiment, as an end-to-end packet loss measurement method, a statistical method is used to predict the original SN decrease phenomenon number from the number of SN decrease phenomena observed by sampling measurement, and the packet Loss is recorded. For this reason, even if it is not possible to detect all the SN lowering phenomenon numbers, it is possible to predict the original SN lowering phenomenon number. Similarly, the original amount when all packets are observed using a statistical method based on the value of the sequence number that has been urged to be retransmitted in the past based on the value observed by sampling. The amount of packet loss generated between the measuring instrument and the receiving terminal is predicted. For this reason, by sampling, even when not all sequence numbers are recorded, it is possible to predict the amount of sequence numbers that have been urged to be retransmitted in the past. Similarly, since the value of the end-to-end packet loss and the value of the packet loss between the measuring instrument and the receiving terminal can be predicted at the time of sampling, the value of the packet loss between the transmitting terminal and the measuring instrument is also predicted. Can do.

  By using these sampling measurement methods, communication quality, “throughput / goodput”, “end-to-end packet loss”, “transmitting terminal—measurement” even in a situation where not all packets of the flow to be measured can be acquired. It is possible to correctly measure “inter-device packet loss” and “measuring device-receiving terminal packet loss”. In addition, since it is not necessary to perform processing on all packets in order to measure the quality of the flow, the measuring instrument does not need high calculation capability.

  In the present embodiment, for the sake of simplification, the description has been given with the apparatus that measures the quality of TCP communication. However, the order of the data sequence is described in the transmission data, and there is a retransmission mechanism for data loss. Is a common technology. Therefore, it includes general protocols having a retransmission mechanism such as BIC-TCP, FastTCP, HSTCP, SCTP, and DCCP.

  As an application area, not only the state of FIG. 1 that does not affect the network to be measured and traffic, but also a format that can acquire data is inserted in the middle between communication terminals, and the network to be measured and traffic are affected. The form shown in FIG. 3 is also possible. The data relay terminal in FIG. 3 is an Ethernet switch that performs data transfer at Layer 2, a router that performs data transfer at Layer 3, a gateway that performs transfer at Layer 4 or higher, or the like. This refers to a terminal that has been transferred with a changed protocol or that has a load balance function or bandwidth control function.

  FIG. 13 shows the application area of the network in the present invention.

  FIG. 15 is a block diagram showing the configuration of the third embodiment of the measuring apparatus 1d according to the present invention.

  In the communication terminal 2b and the communication terminal 3b in the third embodiment, the end-to-end quality is added to the data transmission units 21 and 31 and the data reception units 22 and 32 in the first embodiment and the second embodiment. It consists of measuring units 23 and 33.

The measuring device 1d according to the third embodiment has a probability of one out of every N packets received by the data receiving unit 111, the data receiving unit 112, and the data receiving units 111 and 112, which are data inputs from the branching device. A data preprocessing unit 100a including a sampling processing unit 113 that acquires information, a flow identification unit 114 that performs flow classification using information in the acquired packet, and a flow information storage unit 115 that stores the result. When,
A throughput determination unit 204d that acquires information from the end-to-end quality measurement units 22 and 32 of the communication terminal and acquires an end-to-end throughput; and an end-to-end packet loss that acquires an end-to-end packet loss An end-to-end quality estimation unit 200d composed of a packet loss determination unit 203d,
An SN information management unit 301b that extracts a sequence number (SN) of a passing packet from the flow information storage unit 115, a passing SN storage unit 303b that stores the result, and an ACK number that extracts information from the flow information storage unit 115 A duplicate ACK determination unit 302b that checks the relationship between the continuity and SN, a target event storage unit 304b that stores the number of times that a check target event has occurred, and a packet that occurs between the measuring instrument and the receiving terminal based on that information Measuring device-receiving terminal loss determination unit 305b for calculating the number of losses / rate, and determining the number / rate of transmission terminal-measuring device loss from end-to-end loss information and measuring device-receiving terminal loss information. A packet loss direction specifying unit 300b including a transmission terminal-measuring device loss determination unit 306b,
It comprises a quality result storage unit 400 that stores the result of the end-to-end quality estimation unit 200d and the result of the packet loss direction identification unit 300b.

  Here, the data preprocessing unit 100a, the packet loss direction specifying unit 300b, and the quality result storage unit 400 perform the same configuration / processing as in the first and second embodiments, and thus the description thereof is omitted.

  In the communication terminal 2b in the present embodiment, the data transmission unit 21 performs data transmission, the data reception unit 22 receives data, and the end-to-end communication quality measurement 23 determines the throughput and the end-to-end packet loss. Keep counting. Here, the throughput at the end terminal, the end-to-end packet loss, or the number of packet retransmissions can be counted in many cases with a standard operating system. If the operating system cannot obtain the quality, the quality measurement by observing all standard packets, the throughput of the corresponding flow, and the end-to-end packet loss may be calculated. When the throughput and the end-to-end packet loss are measured, the flow information and the time when the quality is measured are marked, and the result is sent to the measuring device 1d.

  Since the communication terminal 3b in the present embodiment also functions in the same manner as the communication terminal 2b, description thereof is omitted.

  The end-to-end quality estimation unit 200d in the present embodiment performs an end-to-end communication quality analysis at a specified time range or at certain intervals. The throughput determining unit 204d receives the throughput information measured by the end-to-end quality measuring unit 23 or the end-to-end quality measuring unit 33, and sends it to the quality result nucleus unit 400 as data of the corresponding flow information at a corresponding time. Record. Also, the end-to-end loss determination unit 203d receives the end-to-end packet loss information measured by the end-to-end quality measurement unit 23 or the end-to-end quality measurement unit 33, and receives the corresponding flow information. Recorded in the quality result information 400 as data of the corresponding time.

  In the present embodiment, End-to-End quality, throughput and End-to-End packet loss are measured in the communication terminals 2b and 3b. However, End-to-End quality, throughput and End are measured outside the measuring instrument. Any format may be used as long as it measures -to-End packet loss and inputs the result to the measurement apparatus 1d. Measuring the end-to-end quality outside the measuring instrument means measuring with a router or a network analysis device.

  Next, the flow of processing in the measuring device 1d will be described.

  The overall flow of the measuring device 1d is as follows. After the processing of the data preprocessing unit 100a (FIG. 8), the processing of the end-to-end quality estimation unit 200d (based on the information classified / stored there ( 17) and the packet loss direction specifying unit 300b (processing diagram 10) are started.

  Here, the processing of the data pre-processing unit 100a (FIG. 8) and the packet loss direction specifying unit 300b (FIG. 10) perform the same configuration / processing as in the first embodiment, and thus description thereof is omitted.

  Here, FIG. 17 shows an outline of a processing flow in the end-to-end quality estimation unit 200d.

  The processing is started by receiving the End-to-End quality result from the End-to-End quality measurement unit 23 of the communication terminal 2b and the End-to-End quality measurement unit 33 of the communication terminal 3b.

  In the process E-1, End-to-End packet loss information is extracted from the received information, and is recorded in the quality result information storage unit 400 as data of a corresponding time of the corresponding flow information. After this process is completed, the process moves to process E-2.

  In the process E-2, end-to-end throughput information is extracted from the received information, and is recorded in the quality result information storage unit 400 as data of a corresponding time of the corresponding flow information. With this process, the process of the end-to-end quality estimation unit 200d is terminated.

  The processing flow outline is processed every time information is received from the end-to-end quality measurement unit 23 of the communication terminal 2b and the end-to-end quality measurement unit 33 of the communication terminal 3b.

  The above is the processing content of the measuring device 1d according to the third embodiment of the present invention.

  In the conventional technology, when counting the amount of packet loss that occurs between the measuring instrument and the receiving terminal and the packet loss that occurs between the transmitting terminal and the measuring instrument, all packets are observed and packet loss occurs (duplication Every time an ACK occurs), check whether the lost packet has passed in the past, and whether the packet loss occurred between the measuring instrument and the receiving terminal or whether it occurred between the transmitting terminal and the measuring instrument Was judged. For this reason, when this calculation method is applied to the sampled packet group, all duplicate ACKs cannot be observed, and therefore the number of packet losses generated in End-to-End is estimated to be considerably lower than the original value. It will be. Further, when a duplicate ACK occurs, the sequence number passed in the past is confirmed, and if a packet for which retransmission is promoted is lost, the number of phenomena is regarded as a generated packet loss between the measuring instrument and the receiving terminal. This is because it does not consider whether the packet that is being retransmitted really did not pass due to packet loss or could not be observed by sampling, so the amount of packet loss between the measuring instrument and the receiving terminal and the transmitting terminal-measuring instrument The inter-packet loss is classified completely different from the original.

  On the other hand, in this embodiment, the end-to-end communication quality is measured by the communication terminal. Since the communication terminal is in a stage before the packets are aggregated, the processing capability such as CPU, memory, and bus bandwidth necessary for measuring the end-to-end communication quality can be kept low in each terminal. Since the end-to-end communication quality can be obtained at almost the same value regardless of where it is measured in the network, the processing load on the measuring device 1d can be reduced by performing this processing at a terminal other than the measuring device 1d. it can. In addition, based on the value of the sequence number that has been urged to be retransmitted in the past, based on the value observed by sampling, the statistical amount is used to predict the original amount when all packets are observed. The amount of packet loss generated between the measuring instrument and the receiving terminal. For this reason, by sampling, even when not all sequence numbers are recorded, it is possible to predict the amount of sequence numbers that have been urged to be retransmitted in the past. Similarly, since the value of the end-to-end packet loss and the value of the packet loss between the measuring instrument and the receiving terminal can be predicted at the time of sampling, the value of the packet loss between the transmitting terminal and the measuring instrument is also predicted. Can do.

  By using these sampling measurement methods, communication quality, “throughput / goodput”, “end-to-end packet loss”, “transmitting terminal—measurement” even in a situation where not all packets of the flow to be measured can be acquired. It is possible to correctly measure “inter-device packet loss” and “measuring device-receiving terminal packet loss”. In order to measure the quality of these flows, the quality that can be obtained anywhere in the network is measured at a place other than the measuring device, and the quality that can only be measured at this place is measured by the measuring device. By separating the measurement locations, the processing load on the measurement device can be reduced. In addition, in this measuring device, it is not necessary to perform processing on all packets, so that the measuring instrument does not need high calculation capability.

  In the present embodiment, for the sake of simplification, the description has been given with the apparatus that measures the quality of TCP communication. However, the order of the data sequence is described in the transmission data, and there is a retransmission mechanism for data loss. Is a common technology. Therefore, it includes general protocols having a retransmission mechanism such as BIC-TCP, FastTCP, HSTCP, SCTP, and DCCP.

  As an application area, not only the state of FIG. 13 that does not affect the network to be measured and traffic, but also a format capable of acquiring data is inserted in the middle between communication terminals to affect the network to be measured and traffic. The form shown in FIG. 3 is also possible. The data relay terminal in FIG. 3 is an Ethernet switch that performs data transfer at Layer 2, a router that performs data transfer at Layer 3, a gateway that performs transfer at Layer 4 or higher, or the like. This refers to a terminal that has been transferred with a changed protocol or that has a load balance function or bandwidth control function. FIG. 18 is a block diagram showing the configuration of the fourth embodiment of the measuring apparatus 1e according to the present invention.

The measuring device 1e according to the fourth embodiment has a probability of one out of every N packets received by the data receiving unit 111, the data receiving unit 112, and the data receiving units 111 and 112, which are data inputs from the branching device. A data preprocessing unit 100a including a sampling processing unit 113 that acquires information, a flow identification unit 114 that performs flow classification using information in the acquired packet, and a flow information storage unit 115 that stores the result. When,
An end-to-end quality estimation unit 200e for performing end-to-end quality measurement;
An SN information management unit 301c that extracts a sequence number (SN) of a passing packet from the flow information storage unit 115, a passing SN storage unit 303c that stores the result, and information that is extracted from the flow information storage unit 115 A re-SN confirmation unit 302c for checking whether or not the packet sequence number has passed through the measuring device 1e in the past, and a target event storage unit 304c for storing the number of times an event to be checked (re-SN phenomenon) has occurred. A loss determination unit 305c between the measuring device and the receiving terminal that calculates the number / rate of packet loss generated between the measuring device and the receiving terminal based on the information; End-to-End loss information and loss between the measuring device and the receiving terminal A packet loss direction specification composed of a transmission terminal-measuring device loss determination unit 306b that determines the number / rate of transmission terminal-measuring device loss / rate based on information. And 300c,
A quality result storage unit 400 that stores the result of the end-to-end quality estimation unit 200e and the result of the packet loss direction identification unit 300c is configured.

  Here, the data preprocessing unit 100a and the quality result storage unit 400 perform the same configuration / processing as in the first, second, and third embodiments, and thus the description thereof is omitted.

  The end-to-end quality estimation unit 200e here may be any method as long as the end-to-end quality can be calculated. For example, the first embodiment (End-to-End quality estimation unit 200a), the second embodiment (End-to-End quality estimation unit 200b), the third embodiment (End-to -Any end quality estimation unit 200c), or a method of calculating a plurality of these simultaneously and adopting any one of the results may be considered. As an example of the adoption method, a method of adopting a method having a high probability (detection probability) of observing a measurement object in each method is conceivable. The end-to-end quality estimation unit 200a, the end-to-end quality estimation unit 200b, and the end-to-end quality estimation unit 200c perform the same configuration / processing as in the first, second, and third embodiments. Since it does, description is abbreviate | omitted.

  The packet loss direction specifying unit 300c in the present embodiment performs a packet loss direction specifying process at a specified time range or at certain intervals. Here, the start time and the end time of the quality analysis do not necessarily have to completely coincide with the end-to-end quality estimation unit 200e. The SN information management unit 301c extracts the data-side packet of the corresponding flow from the flow information storage unit 115, and stores the existing sequence number in the passing SN storage unit 303c. At this time, record each sequence number and how many times it passes. The re-SN confirmation unit 302c finds a sequence number in which the number of passages is two or more from the passage SN storage unit 303b, and records the number in the target event number recording unit 304c. When the processing in the SN information management unit 301c and the processing in the re-SN confirmation unit 302c are completed for all packets subject to quality analysis, the processing in the loss determination unit 305c between the measuring instrument and the receiving terminal and the loss between the transmitting terminal and the measuring device Processing in the determination unit 306b is performed. The measuring instrument-receiving terminal loss determination unit 305c obtains the information of the target event number recording unit 304c of the corresponding flow, and calculates the number of losses between the measuring instrument-receiving terminal by calculating equation (16).

Here, D is the number of target events in the present embodiment, and s is the sampling probability. This calculates the number at which the sequence number is observed more than once (see FIG. 19). The fact that the sequence number is observed twice means that the first-pass packet has been lost and that loss has occurred between the measuring instrument and the receiving terminal. From the probability that the number can be observed under the rate, the original loss between the measuring instrument and the receiving terminal is calculated.

  After calculating the packet loss between the measuring instrument and the receiving terminal by the equation (16), the packet loss between the transmitting terminal and the measuring instrument can be calculated by subtracting the value from the end-to-end packet loss.

  Next, with reference to FIG. 20, the flow of processing in the measuring apparatus 1e will be described.

  The overall flow of the measuring device 1e is as follows. After the processing of the data preprocessing unit 100a (FIG. 8) is performed, the end-to-end quality estimation unit 200e and the packet are processed based on the information classified / stored there. The process of the loss direction identification unit 300b is started.

  Here, the processing of the data preprocessing unit 100a (FIG. 7) performs the same configuration / processing as in the first embodiment, and thus description thereof is omitted.

  The processing of the end-to-end quality estimation unit 200e may be any method as long as the end-to-end quality can be calculated. For example, the first embodiment (End-to-End quality estimation unit 200a), the second embodiment (End-to-End quality estimation unit 200b), the third embodiment (End-to -Any end quality estimation unit 200c), or a method of calculating a plurality of these simultaneously and adopting any one of the results may be considered. The end-to-end quality estimation unit 200a, the end-to-end quality estimation unit 200b, and the end-to-end quality estimation unit 200c perform the same configuration / processing as in the first, second, and third embodiments. Since it does, description is abbreviate | omitted.

  FIG. 20 shows an outline of a processing flow in the packet loss direction specifying unit 300c.

  Processing is started by fetching only the Data side of the quality measurement target flow from the flow information storage unit 115 into the packet loss direction specifying unit 300c. This process is process F-1. After this process is completed, the process moves to process F-2.

  In the process F-2, the SN information management unit 301c manages sequence numbers. If the acquired packet is the ACK side information of the corresponding flow, the process moves to process F-4 without performing any process. If the packet is on the data side of the flow, the process moves to process F-3.

  In the process F-3, the sequence number of the packet acquired by the SN information management unit 301c is recorded in the passing SN storage unit 303c. If the corresponding sequence number has passed in the past, information on how many times it has passed is additionally written, and the process proceeds to process F-4.

  In process F-4, the process of the re-SN confirmation unit 302c is performed. Confirming with the re-SN confirms the passing sequence numbers in order. If the sequence number does not decrease, the process moves to process F-7. When the sequence number is lowered, the process proceeds to process F-5.

  In process F-5, the re-SN confirmation unit 302c checks the passing SN management unit 303c to check whether the current sequence number is observed for the first time or the second time or later. If it is the first observation, the process proceeds to process F-7, and if it is the second and subsequent passes, the process proceeds to process F-6.

  In process F-6, the target event number of the flow to which the current packet belongs is read from the target event number recording unit 304c, counted up, and the updated information is stored again. After this processing, the process moves to F-7.

  In the process F-7, it is confirmed whether or not all the packets in the observation period target period have been read by the SN information management unit 301c and the re-SN confirmation unit 303b. If there is a packet that has not yet been determined, the next packet is read, and the process returns to process F- 1 in order to check the passage of the SN number and update the target event count. If all packets have been determined, the process moves to process F-8 in order to calculate quality information during that period.

  In process F-8, the measuring instrument-receiving terminal loss determination unit 305c calculates the measuring instrument-receiving terminal packet loss. The number of packet loss between the measuring instrument and the receiving terminal can be calculated from the probability of observing both the packet retransmission packet (normal transmission and transmission at the time of retransmission of the packet) from the sampling probability. As an example of the calculation method, there is a formula (16). After this process is completed, the process moves to process F-9.

  In the process F-9, the transmission terminal-measuring device loss determination unit 306b calculates the number of packet loss between the transmitting terminal and the measuring device by calculating the equation (9). With this process, the process of the packet loss direction determination unit 300b is terminated.

  The above is the processing content of the measuring device 1e according to the fourth embodiment of the present invention.

  In the conventional technology, when counting the amount of packet loss that occurs between the measuring instrument and the receiving terminal and the packet loss that occurs between the transmitting terminal and the measuring instrument, all packets are observed and packet loss occurs (duplication Every time an ACK occurs), check whether the lost packet has passed in the past, and whether the packet loss occurred between the measuring instrument and the receiving terminal or whether it occurred between the transmitting terminal and the measuring instrument Was judged. For this reason, when this calculation method is applied to the sampled packet group, all duplicate ACKs cannot be observed, and therefore the number of packet losses generated in End-to-End is estimated to be considerably lower than the original value. It will be. Further, when a duplicate ACK occurs, the sequence number passed in the past is confirmed, and if a packet for which retransmission is promoted is lost, the number of phenomena is regarded as a generated packet loss between the measuring instrument and the receiving terminal. This is because it does not consider whether the packet that is being retransmitted really did not pass due to packet loss or could not be observed by sampling, so the amount of packet loss between the measuring instrument and the receiving terminal and the transmitting terminal-measuring instrument The inter-packet loss is classified completely different from the original.

  On the other hand, in the present embodiment, by calculating the number of retransmission phenomena observed during sampling measurement and the probability of observable retransmission phenomena, the number of retransmission phenomena that would have occurred originally is calculated. Therefore, by sampling, it is possible to predict the original number of retransmission phenomena even when all sequence numbers are not recorded. Similarly, since the value of the end-to-end packet loss and the value of the packet loss between the measuring instrument and the receiving terminal can be predicted at the time of sampling, the value of the packet loss between the transmitting terminal and the measuring instrument is also predicted. Can do. Further, the end-to-end quality estimation unit 200e of the present embodiment is sampled when the end-to-end communication quality is measured only on the Data side as in the end-to-end quality estimation unit 200c of the second embodiment. All the communication qualities can be measured with only the Data side packet, and the quality can be measured with a smaller number of packets.

  By using these sampling measurement methods, communication quality, “throughput / goodput”, “end-to-end packet loss”, “transmitting terminal—measurement” even in a situation where not all packets of the flow to be measured can be acquired. It is possible to correctly measure “inter-device packet loss” and “measuring device-receiving terminal packet loss”. In addition, since it is not necessary to perform processing on all packets in order to measure the quality of the flow, the measuring instrument does not need high calculation capability.

  In the present embodiment, for the sake of simplification, the description has been given with the apparatus that measures the quality of TCP communication. However, the order of the data sequence is described in the transmission data, and there is a retransmission mechanism for data loss. Is a common technology. Therefore, it includes general protocols having a retransmission mechanism such as BIC-TCP, FastTCP, HSTCP, SCTP, and DCCP.

  As an application area, not only the state of FIG. 1 that does not affect the network to be measured and traffic, but also a format that can acquire data is inserted in the middle between communication terminals, and the network to be measured and traffic are affected. The form shown in FIG. 3 is also possible. The data relay terminal 6 in FIG. 3 is an Ethernet switch that performs data transfer at Layer 2, a router that performs data transfer at Layer 3, a gateway that performs transfer at Layer 4 or higher, or the like. This refers to a terminal to which a protocol is changed or transferred, or a load balance function or bandwidth control function is added.

  FIG. 21 is a block diagram showing the configuration of the fifth embodiment of the measuring apparatus 1f according to the present invention.

The measuring device 1f according to the fifth embodiment includes a data reception unit 111 and a data reception unit 112 that are data inputs from the branching device, and packets that are input by the data reception units 111 and 112 with a probability of 1 in N packets. A data preprocessing unit 100a including a sampling processing unit 113 that acquires information, a flow identification unit 114 that performs flow classification using information in the acquired packet, and a flow information storage unit 115 that stores the result. When,
An end-to-end quality estimation unit 200e for performing end-to-end quality measurement;
A packet loss direction specifying unit 300c that specifies the packet loss direction from only the Data side packet, a packet loss direction specifying unit 300b that specifies the packet loss direction using both the Data side packet and the ACK side packet, and the two packet losses. A packet composed of an adoption loss determination unit 307d that determines which one of the results of the direction identification unit is adopted, and a detection probability calculation unit 308d that obtains information from the quality result storage unit 400 and calculates a detection probability. A loss direction identification unit 300d;
A quality result storage unit 400 that stores the result of the end-to-end quality estimation unit 200e and the result of the packet loss direction specification unit 300d is configured.

  Here, the data preprocessing unit 100a and the quality result storage unit 400 perform the same configuration / processing as in the first, second, and third embodiments, and thus the description thereof is omitted.

  The end-to-end quality estimation unit 200e here may be any method as long as the end-to-end quality can be calculated. For example, the first embodiment (End-to-End quality estimation unit 200a), the second embodiment (End-to-End quality estimation unit 200b), the third embodiment (End-to -Any end quality estimation unit 200c), or a method of calculating a plurality of these simultaneously and adopting any one of the results may be considered. As an example of the adoption method, a method of adopting a method having a high probability (detection probability) of observing a measurement object in each method may be considered. The end-to-end quality estimation unit 200a, the end-to-end quality estimation unit 200b, and the end-to-end quality estimation unit 200c perform the same configuration / processing as in the first, second, and third embodiments. Since it does, description is abbreviate | omitted.

  Further, since the packet loss direction specifying unit 300b performs the same configuration / processing as in the first, second, and third embodiments, the packet loss direction specifying unit 300c has the same configuration / processing as in the fourth embodiment. Since the process is performed, the description is omitted.

  The packet loss direction specifying unit 300d in the present embodiment performs the packet loss direction specifying process at a specified time range or at certain intervals. Here, the start time and the end time of the quality analysis do not necessarily have to completely coincide with the end-to-end quality estimation unit 200e. The packet loss direction identification unit 300b extracts the ACK side packet and the Data side packet of the corresponding flow from the flow information storage unit 115, and calculates “transmission terminal-measuring device packet loss” and “measuring device-receiving terminal packet loss”. At the same time, the packet loss direction specifying unit 300c also extracts the data-side packet of the corresponding flow from the flow information storage unit 115 and calculates "packet loss between the transmitting terminal and measuring device" and "packet loss between measuring device and receiving terminal". Next, the detection probability calculation unit 308d determines the detection probabilities of the packet loss direction specification unit 300b and the packet loss direction specification unit 300c from the End-to-End packet loss obtained from the quality result storage unit 400 and the sampling probability that is a set value. Perform the calculation. The detection probability of the packet loss direction specifying unit 300b is the denominator of the equation (5). Alternatively, it is the value of the detection probability paired with the current End-to-End packet loss rate in the detection probability table 2. The detection probability of the packet loss direction specifying unit 300c is the denominator of the equation (16). It is the square of the sampling probability. After this processing, the adopted loss determination unit 307d compares the detection probability of the packet loss direction specifying unit 300b with the detection probability of the packet loss direction specifying unit 300c, and transmits the communication quality result calculated by the method having a large detection probability as “transmission”. The packet loss between the terminal and the measuring device and the packet loss between the measuring device and the receiving terminal are adopted and stored in the quality result storage unit 400.

  Next, with reference to FIG. 22, the flow of processing in the measuring device 1f will be described.

  The overall flow of the measuring device 1f is as follows. After the processing of the data preprocessing unit 100a (FIG. 8) is performed, the end-to-end quality estimation unit 200e The process of the loss direction identification unit 300b is started.

  Here, the processing (FIG. 8) of the data preprocessing unit 100a performs the same configuration / processing as in the first embodiment, and thus description thereof is omitted.

  The processing of the end-to-end quality estimation unit 200e may be any method as long as the end-to-end quality can be calculated. For example, the first embodiment (End-to-End quality estimation unit 200a), the second embodiment (End-to-End quality estimation unit 200b), the third embodiment (End-to -Any end quality estimation unit 200c), or a method of calculating a plurality of these simultaneously and adopting any one of the results may be considered. The end-to-end quality estimation unit 200a, the end-to-end quality estimation unit 200b, and the end-to-end quality estimation unit 200c perform the same configuration / processing as in the first, second, and third embodiments. Since it does, description is abbreviate | omitted.

  Further, since the packet loss direction specifying unit 300b performs the same configuration / processing as in the first, second, and third embodiments, the packet loss direction specifying unit 300c has the same configuration / processing as in the fourth embodiment. Since the process is performed, the description is omitted.

  FIG. 22 shows an outline of a processing flow in the packet loss direction specifying unit 300d.

  The processing is started by fetching the Data side packet and the ACK side packet of the quality measurement target flow from the flow information storage unit 115 into the packet loss direction specifying unit 300b. With this processing, it is possible to obtain “packet loss between transmitting terminal and measuring device” and “packet loss between measuring device and receiving terminal” during a certain measurement period. This process is process G-1. After this process is completed, the process moves to process G-2.

  In the process G-2, the process is started by fetching the Data side packet of the quality measurement target flow from the flow information storage unit 115 into the packet loss direction specifying unit 300c. With this processing, it is possible to obtain “packet loss between transmitting terminal and measuring device” and “packet loss between measuring device and receiving terminal” during a certain measurement period. After this process is completed, the process moves to process G-3.

  In the process G-3, the end-to-end packet loss and the sampling probability are input from the quality result storage unit 400 to the detection probability calculation unit 308d, and the detection probability of the packet loss direction specification unit 300b and the detection of the packet loss direction specification unit 300c. Calculate the probability. After the calculation process, the process proceeds to process G-4.

  In process G-4, each detection probability calculated in process G-3 is compared by the adoption loss determination unit 307d. If the detection probability of the packet loss direction identification unit 300b is larger, go to process G-5. If the detection probability of the packet loss direction identification unit 300c is large, the process proceeds to process G-6.

  In the process G-5, the adopted loss determination unit 307d determines that the calculation result of the packet loss direction specifying unit 300b is the final “transmission terminal-measuring device packet loss” and “measuring device-receiving terminal packet loss”. Then, the result is recorded in the quality result storage unit 400, and measurement of the current observation period is finished.

  In process G-6, the adopted loss determination unit 307d determines that the calculation result of the packet loss direction identification unit 300c is the final “transmission terminal-measuring device packet loss” and “measuring device-receiving terminal packet loss”. Then, the result is recorded in the quality result storage unit 400, and measurement of the current observation period is finished.

  The above is the processing content of the measuring device 1f according to the fifth embodiment of the present invention.

  In the conventional technology, when counting the amount of packet loss that occurs between the measuring instrument and the receiving terminal and the packet loss that occurs between the transmitting terminal and the measuring instrument, all packets are observed and packet loss occurs (duplication Every time an ACK occurs), check whether the lost packet has passed in the past, and whether the packet loss occurred between the measuring instrument and the receiving terminal or whether it occurred between the transmitting terminal and the measuring instrument Was judged. For this reason, when this calculation method is applied to the sampled packet group, all duplicate ACKs cannot be observed, and therefore the number of packet losses generated in End-to-End is estimated to be considerably lower than the original value. It will be. Further, when a duplicate ACK occurs, the sequence number passed in the past is confirmed, and if a packet for which retransmission is promoted is lost, the number of phenomena is regarded as a generated packet loss between the measuring instrument and the receiving terminal. This is because it does not consider whether the packet that is being retransmitted really did not pass due to packet loss or could not be observed by sampling, so the amount of packet loss between the measuring instrument and the receiving terminal and the transmitting terminal-measuring instrument The inter-packet loss is classified completely different from the original.

  On the other hand, in the present embodiment, the original communication quality “throughput / goodput”, “end-to-end packet loss”, “transmitting terminal—measurement” is obtained by obtaining the detection probability of each observed event even during sampling measurement. It is possible to correctly measure “inter-device packet loss” and “measuring device-receiving terminal packet loss”. Furthermore, the accuracy of the estimation result of communication quality can be improved by calculating each quality by a plurality of methods and adopting a method with a high detection probability. In addition, since it is not necessary to perform processing on all packets in order to measure the quality of the flow, the measuring instrument does not need high calculation capability.

  In the present embodiment, for the sake of simplification, the description has been given with the apparatus that measures the quality of TCP communication. However, the order of the data sequence is described in the transmission data, and there is a retransmission mechanism for data loss. Is a common technology. Therefore, it includes general protocols having a retransmission mechanism such as BIC-TCP, FastTCP, HSTCP, SCTP, and DCCP.

  As an application area, not only the state of FIG. 1 that does not affect the network to be measured and traffic, but also a format that can acquire data is inserted in the middle between communication terminals, and the network to be measured and traffic are affected. The form shown in FIG. 3 is also possible. The data relay terminal 6 in FIG. 3 is an Ethernet switch that performs data transfer at Layer 2, a router that performs data transfer at Layer 3, a gateway that performs transfer at Layer 4 or higher, or the like. This refers to a terminal to which a protocol is changed or transferred, or a load balance function or bandwidth control function is added.

  The sixth embodiment includes a measurement server that measures network communication quality with a plurality of measurement devices and collects the results.

  The measurement device 1111, the measurement device 1112, and the measurement device 111 </ b> N in FIG. 23 perform any of the operations from the measurement device 1 b to the measurement device 1 f of the first to fifth embodiments, and perform communication at each measurement position. Quality “throughput / goodput”, “end-to-end packet loss”, “packet loss between transmitting terminal and measuring device”, and “packet loss between measuring device and receiving terminal” are measured. The operation of the measuring device here is the same as that of any one of the first to fifth embodiments or a combination thereof, and the description thereof will be omitted.

  Each measuring device adds the measurement period, time, relevant flow information, and measurement position information to the measured communication quality, and transmits it to the measurement server.

  FIG. 24 is a block diagram showing the configuration of the measurement server 2000 according to the sixth embodiment of the present invention.

  The measurement server 2000 includes a quality result classification unit 2001, each measurement point loss recording unit 2002, a measurement period loss calculation unit 2003, and an inter-instrument loss recording unit 2004.

  The communication quality result input from each measurement device is received by the quality result classification unit 2001, classified for each flow, period, measurement point, and time information, and recorded in each measurement point loss recording unit 2002. The inter-instrument loss calculation unit 2003 acquires the quality at each measurement point at the similar time of the same flow information, and calculates the inter-measuring device packet loss by the following equation (17) or (18). This recording is performed for each measuring instrument, and the result is recorded in the inter-measuring loss recording unit 2004.

Packet loss between measuring device A and measuring device B = | ("Transmission terminal-measuring device packet loss" of measuring device A)
− (“Transmission terminal-measuring device packet loss” of measuring device B) | (17)
Packet loss between measuring device A and measuring device B = | ("Measuring device-receiving terminal packet loss" of measuring device A)
-("Measurement Device-Receiving Terminal Packet Loss" of Measurement Device B) | (18)
Next, the flow of processing in the measurement server 2000 will be described with reference to FIG.

  As an overall flow of the measurement system, the communication quality is measured by each measurement device, and the result is input to the measurement server 2000, and the process is started.

  Here, since each measuring apparatus performs the same configuration / process as any one of the first to fifth embodiments or a combination thereof, description thereof is omitted.

  FIG. 25 shows an outline of a processing flow in the measurement server 2000.

  The communication quality “throughput / goodput”, “End-to-End packet loss”, “transmitting terminal-measuring device packet loss”, and “measuring device-receiving terminal packet loss” calculated by each measuring device are classified into quality results. The data is received by the unit 2001, classified by time, measurement interval, and belonging flow, and the result is recorded in each measurement point loss recording unit 2002. This is processing H-1. The process proceeds to post-process H-2.

  In the process H-2, the inter-instrument loss calculation unit 2003 calculates the inter-instrument packet loss using the equations (17) and (18). The result is recorded in the inter-instrument loss recording unit 2004. When “inter-measuring packet loss” is calculated among all the measuring instruments, the process is terminated.

  This series of processing is performed on all accumulated data every time a communication result is input from each measuring device.

  The above is the processing contents of the measurement server in the sixth embodiment of the present invention.

  In the conventional technology, it is necessary to observe all the packets when counting the amount of packet loss generated between the measuring instrument and the receiving terminal and the packet loss occurring between the transmitting terminal and the measuring instrument. Processing load was heavy. Thus, in order to increase the number of observation points, a measuring device with high processing performance is required, and it is difficult to take many observation points. As a result, the inter-instrument packet loss cannot be measured with a fine granularity, and it is only possible to roughly know in which section the packet loss has occurred.

  On the other hand, in this embodiment, the communication quality “throughput / goodput”, “End-to-End packet loss”, “transmission terminal-measuring device packet loss”, “measuring device-receiving terminal packet loss” by sampling measurement. "Can be calculated correctly, the processing load of one measuring device is smaller than before. As a result, it is possible to take many points as observation points, and it is possible to specify a section where packet loss has occurred with a finer system than before.

  In the present embodiment, for the sake of simplification, the description has been given with the apparatus that measures the quality of TCP communication. However, the order of the data sequence is described in the transmission data, and there is a retransmission mechanism for data loss. Is a common technology. Therefore, it includes general protocols having a retransmission mechanism such as BIC-TCP, FastTCP, HSTCP, SCTP, and DCCP.

  Further, as an application area, as a method of passively acquiring network packets, not only the state of FIG. 23 that does not affect the network to be measured and traffic, but also a format between which data can be acquired, as long as it is in the middle between communication terminals The configuration shown in FIG. 3 is also possible, which is inserted into the network and affects the network to be measured and the traffic. The data relay terminal 6 in FIG. 3 is an Ethernet switch that performs data transfer at Layer 2, a router that performs data transfer at Layer 3, a gateway that performs transfer at Layer 4 or higher, or the like. This refers to a terminal to which a protocol is changed or transferred, or a load balance function or bandwidth control function is added.

  The first effect of the present embodiment is that the communication quality measurement system does not require high calculation capability.

  The reason for this is that, according to the present embodiment, an index relating to communication quality can be calculated simply by acquiring a part of a packet sequence, so that the number of packets to be calculated in the quality measurement system can be greatly reduced. It is.

  The second effect of the present embodiment is that the quality measurement system can accurately estimate an index related to communication quality even in a situation where all packets cannot be acquired.

  This is because according to the present embodiment, an index relating to communication quality can be calculated only by acquiring a part of the packet sequence.

  The third effect of the present embodiment is that the section in which the packet loss has occurred can be accurately determined by the quality measurement system.

  The reason for this is that, according to this embodiment, the processing capability required by one measuring device is reduced by this embodiment, so that measuring devices can be deployed in many places in the network, and the section where packet loss occurs can be more detailed. It is because it becomes possible to specify.

It is a figure which shows the application area | region of this technique. It is a block diagram of the conventional method. It is a figure which shows the application area | region of the 2nd method of this technique. It is a block diagram with the measuring device in a first embodiment. It is a figure which shows the event observed in the End-to-End quality estimation part 200b. It is a figure which shows the relationship between the duplication degree of confirmation response signal, and frequency. It is a figure which shows the event observed in the packet loss direction specific | specification part 300b. It is a figure which shows the outline | summary of the processing flow in the pre-processing part in the measuring device in 1st embodiment. It is a figure which shows the outline | summary of the processing flow in the End-to-End quality measurement part in the measuring device in 1st embodiment. It is a figure which shows the outline | summary of the processing flow in the packet loss direction specific | specification part in the measuring device in 1st embodiment. It is a block diagram with the measuring device in a second embodiment. It is a figure which shows the event observed in the End-to-End quality estimation part 200c. It is a figure which shows the relationship between the fall number of a transmission order number, and frequency. It is a figure which shows the outline | summary of the processing flow in the End-to-End quality measurement part in the measuring device in 2nd embodiment. It is a figure which shows the application area | region of the 3rd method of this technique. It is a block diagram with the measuring device in a third embodiment. It is a figure which shows the outline | summary of the processing flow in the End-to-End quality measurement part in the measuring device in 3rd embodiment. It is a block diagram with the measuring device in a fourth embodiment. It is a figure which shows the event observed in the packet loss direction specific | specification part 300c. It is a figure which shows the outline | summary of the processing flow in the packet loss direction specific | specification part in the measuring device in 4th embodiment. It is a block diagram with the measuring device in a fifth embodiment. It is a figure which shows the outline | summary of the processing flow in the packet loss direction specific | specification part in the measuring device in 5th embodiment. It is a figure showing an application field of the 4th method of this art. It is a block diagram with the measuring device in a 6th embodiment. It is a figure which shows the outline | summary of the processing flow in the measurement server in 6th embodiment.

Explanation of symbols

1, 1111, 1112, 111N Measuring device 2a, 3a Communication terminal 4, 5, 41, 42, 4N, 51, 52, 5N Branch device 6 Data relay terminal 21, 32 Data transmitting unit 22, 31 Data receiving unit 23, 33 End-to-End Quality Measurement Unit 2000 Measurement Server

Claims (24)

A network quality measurement method for measuring network quality,
Acquired the data transmission sequence number of some transmission data sent from the data transmission side to the data reception side and the reception confirmation number of some acknowledgment signals sent from the data reception side to the data transmission side. Based on the result, the step of calculating at least one of the number of data loss or the data loss rate occurring between the measurement point and the data receiving terminal or the number of data loss or the data loss rate occurring between the data transmitting terminal and the measurement point is performed. A network quality measuring method comprising: A network quality measuring method for measuring the quality of the network according to claim 1,
Based on the result of the previous period confirmation, the measurement point and the data based on the step of confirming whether the acquired confirmation response number is the same number or more of the number of times specified in advance, the step of confirming whether or not the data transmission order number has passed A network quality characterized by calculating at least one of the number of data loss or data loss rate occurring between the receiving terminals or the number of data loss times or data loss rate occurring between the data transmitting terminal and the measurement point Measurement method. A network quality measuring method for measuring the quality of a network according to claim 1 or 2,
A network quality measurement method characterized in that, as a step of confirming whether or not a data transmission sequence number has passed, a data transmission sequence number corresponding to a number instructed to be retransmitted by an acknowledgment signal is targeted. A network quality measuring method for measuring the quality of a network according to any one of claims 1 to 3,
The number of data loss that occurred between the measurement point and the data receiving terminal from the step of confirming whether the acquired confirmation response number continues more than the number of times specified in advance and the step of confirming whether the data transmission sequence number has passed Alternatively, in the step of calculating one or more of the data loss rate or the number of data loss or the data loss rate that occurred between the data transmission terminal and the measurement point,
A network quality measurement method characterized by using one or more of data loss count, data loss rate, goodput, sampling probability, or duplication degree designated for confirmation of duplication of acknowledgment number as a parameter. A network quality measurement method for measuring network quality,
Acquire the data transmission sequence number of some transmission data sent from the data transmission side to the data reception side, and based on the results obtained in the previous period, the number of data loss or data loss rate or data that occurred between the measurement point and the data receiving terminal A network quality measurement method comprising a step of calculating at least one of the number of times of data loss and the data loss rate occurring between a transmission terminal and a measurement point. A method for measuring network quality according to any one of claims 1 to 5,
The detection probability calculated by the means for calculating the detection probability of the counted event is
A network quality measurement method characterized by the following equation.

Here, s is a sampling probability which is a data observation probability, Th is the number of data communications generated between the transmitting terminal and the receiving terminal, P is a number of data loss occurring between the transmitting terminal and the receiving terminal, and k is a transmission order number. The number of consecutive acknowledgment signal numbers, which is the confirmation condition, and b is a constant. A network quality measuring method for measuring the quality of a network according to claim 5,
In the acquired transmission sequence number, the step of counting the number of times the specified transmission sequence number is the same or more than the specified number, and the number of data loss or data that occurred between the measurement point and the data receiving terminal based on the previous measurement count A network quality measurement method comprising a step of calculating at least one of a loss rate or the number of times of data loss occurring between a data transmission terminal and a measurement point or a data loss rate. A network quality measuring method for measuring the quality of a network according to claim 5 or 7,
The number of data loss or data loss rate between the measurement point and the data receiving terminal or the data transmission terminal and the measurement point In the step of calculating one or more of the number of data loss or the data loss rate that occurred between
A network quality measurement method using sampling probability. The method for measuring network quality according to any one of claims 1 to 8, wherein the number of times that the transmission order number is equal to or more than the specified number that has been counted, or a sequence number that is instructed to be retransmitted. From the number of passes
Detection probability used to calculate one or more of the number of data loss or data loss rate that occurred between the measurement point and the data receiving terminal or the number of data loss or data loss rate that occurred between the data transmission terminal and the measurement point The
A network quality measurement method characterized in that a packet loss rate, a sampling probability, or a combination of both and a detection probability are acquired by searching from a database stored in association with each other. A network quality measurement method for measuring network quality,
Acquired the data transmission order number of some transmission data sent from the data transmission side to the data reception side and the reception confirmation number of some acknowledgment signals sent from the data reception side to the data transmission side. Based on the result, the step of calculating at least one of the number of data loss or the data loss rate occurring between the measurement point and the data receiving terminal or the number of data loss or the data loss rate occurring between the data transmitting terminal and the measurement point is performed. A network quality measuring method characterized by comprising a plurality. A method for measuring network quality according to claim 10, comprising:
A network quality measurement method characterized by adopting a means having the highest detection probability among a plurality of calculated detection probabilities. A method for measuring network quality according to any one of claims 1 to 11,
By installing multiple observation points that measure network quality in the network, the number of data loss or data loss rate that occurred between the measurement point calculated at each point and the data receiving terminal or the data transmission terminal and the measurement point occurred A network quality measurement method comprising a step of calculating the number of data loss or the data loss rate between each measurement point by collecting at least one of the number of data loss and the data loss rate. A network quality measuring device for measuring network quality,
Acquired the data transmission sequence number of some transmission data sent from the data transmission side to the data reception side and the reception confirmation number of some acknowledgment signals sent from the data reception side to the data transmission side. Based on the result, means for calculating at least one of the number of data loss or data loss rate occurring between the measurement point and the data receiving terminal or the number of data loss times or data loss rate occurring between the data transmitting terminal and the measurement point. A network quality measuring device comprising: A network quality measuring device for measuring the quality of a network according to claim 13,
A means for confirming whether or not the same number of confirmation response numbers acquired continues for the number of times specified in advance, a means for confirming whether or not the data transmission sequence number has passed, and a measurement point and data based on the previous confirmation result Network quality characterized by having means for calculating at least one of the number of data loss or data loss rate occurring between the receiving terminals or the number of data loss times or data loss rate occurring between the data transmitting terminal and the measurement point Measuring device. A network quality measuring device for measuring the quality of a network according to claim 13 or 14,
A network quality measuring apparatus characterized in that, as means for confirming whether or not a data transmission order number has passed, a data transmission order number corresponding to a number for which retransmission is instructed by an acknowledgment signal is targeted. A network quality measuring device for measuring the quality of a network according to any one of claims 13 to 15,
The number of data loss that occurred between the measurement point and the data receiving terminal from the means to check whether the acquired acknowledgment number has continued more than the number of times specified in advance and the means to check whether the data transmission sequence number has passed or not Alternatively, in the means for calculating at least one of the data loss rate or the number of data loss times or data loss rate that occurred between the data transmission terminal and the measurement point,
A network quality measuring apparatus using one or more of data loss count, data loss rate, goodput, sampling probability, or duplication degree specified for confirmation of duplication of acknowledgment number as a parameter. A network quality measuring device for measuring network quality,
Acquire the data transmission sequence number of some transmission data sent from the data transmission side to the data reception side, and based on the results obtained in the previous period, the number of data loss or data loss rate or data that occurred between the measurement point and the data receiving terminal A network quality measuring apparatus comprising means for calculating at least one of the number of data loss and the data loss rate occurring between a transmitting terminal and a measuring point. An apparatus for measuring network quality according to any one of claims 13 to 17,
The detection probability calculated by the means for calculating the detection probability of the counted event is
A network quality measuring device characterized by the following equation.

Here, s is a sampling probability which is a data observation probability, Th is the number of data communications generated between the transmitting terminal and the receiving terminal, P is a number of data loss occurring between the transmitting terminal and the receiving terminal, and k is a transmission order number. The number of consecutive acknowledgment signal numbers, which is the confirmation condition, and b is a constant. A network quality measuring device for measuring the quality of a network according to claim 17,
In the acquired transmission order number, the number of times the transmission order number is the same as the specified number or more and the number of data loss or data that occurred between the measurement point and the data receiving terminal based on the previous measurement count A network quality measuring device comprising means for calculating one or more of a loss rate or the number of data loss times or data loss rates occurring between a data transmission terminal and a measurement point. A network quality measuring device for measuring the quality of a network according to claim 17 or 19,
The number of data loss or data loss rate between the measurement point and the data receiving terminal or the data transmission terminal and the measurement point In the means of calculating one or more of the number of data loss or the data loss rate that occurred between
A network quality measuring apparatus using a sampling probability. The apparatus for measuring network quality according to any one of claims 13 to 20, wherein the number of times that the transmission order number is equal to or more than the specified specified number counted or the sequence number for which retransmission is instructed From the number of passes
Detection probability used to calculate one or more of the number of data loss or data loss rate that occurred between the measurement point and the data receiving terminal or the number of data loss or data loss rate that occurred between the data transmission terminal and the measurement point The
A network quality measuring apparatus, wherein a packet loss rate, a sampling probability, or a combination of both are obtained by searching from a database stored in association with a detection probability. A network quality measuring device for measuring network quality,
Acquired the data transmission sequence number of some transmission data sent from the data transmission side to the data reception side and the reception confirmation number of some acknowledgment signals sent from the data reception side to the data transmission side. Based on the result, means for calculating at least one of the number of data loss or data loss rate occurring between the measurement point and the data receiving terminal or the number of data loss times or data loss rate occurring between the data transmitting terminal and the measurement point. A network quality measuring device having a plurality of network quality measuring devices. An apparatus for measuring network quality according to claim 22,
A network quality measuring apparatus characterized by adopting a means having the highest detection probability among a plurality of calculated detection probabilities. An apparatus for measuring network quality according to any one of claims 13 to 23, comprising:
By installing multiple observation points that measure network quality in the network, the number of data loss or data loss rate that occurred between the measurement point calculated at each point and the data receiving terminal or the data transmission terminal and the measurement point occurred A network quality measuring apparatus comprising means for calculating the number of data loss or the data loss rate between each measurement point by collecting at least one of the number of data loss and the data loss rate.

Images