JP4280988B2 - Network quality evaluation measurement method and network quality evaluation apparatus - Google Patents

Description

  The present invention relates to a network quality evaluation measurement method and a network quality evaluation apparatus, and more specifically, in a network that transfers data in real time based on a protocol including a sequence number such as that using RTP (Real-Time Transport Protocol). The present invention relates to an improvement of a network quality evaluation measurement method and a network quality evaluation apparatus to be used.

  In paragraph 0019 of Patent Document 1, it is described that RTP is used in real-time communication as one of UDP (User Datagram Protocol).

  The present invention relates to continuous measurement of packet loss (missing) in data transfer based on a protocol including a sequence number using RTP. Patent Document 1 relates to continuous measurement of packet loss as in the present invention. There is no specific description.

JP 2003-298560 A

  The Internet is a familiar communication network. On the Internet, not only text information but also voice and image information can be exchanged as data, and its services include data transmission services such as e-mail, net news, file transfer (ftp), remote login (telnet), Gopher, There are also information retrieval services such as WWW and real-time voice communication services such as Internet telephone (VoIP). According to the Internet, information can be exchanged instantly without being restricted by companies and countries, so it is widely used in various fields such as individuals, companies, government offices, educational institutions, and research institutions. Yes.

  In such an Internet, if the communication quality of the communication network does not meet the quality required by the service you want to use, packet loss, delay time fluctuation, etc. occur, and in the case of voice information, the received voice is interrupted In the case of moving image information, the motion of the received moving image may become unnatural.

  However, in the Internet, a route along the communication network is shared by a plurality of terminals and services, so if the number of users accessing at the same time increases, the communication quality will be lower than when the number of users is small. In addition, although the bandwidth of the line constituting the communication network is generally designed to be wide in the main line system part and narrow in the access system part, the access system part may be overloaded when actually operated.

  In addition, the processing speed, buffer length, processing method, etc. of connected devices such as routers have a great influence on network characteristics.

  In any case, Internet service providers must provide users with a more stable service from the standpoint of providing a part of the communication network for a fee, and if the quality varies from service to service, each cause Must be analyzed accurately.

  It is also desirable for the user to be able to accurately grasp the quality of various services in real time in order to determine whether the quality of various services used on the Internet is commensurate with the communication network usage fee.

  By the way, as a protocol for transferring data such as audio and video in real time, there is RTP defined in RFC (Request For Comment) 1889 and RFC 1890. RTP alone does not secure resources or guarantee quality of service (QOS), and does not select transport layer or network layer protocols, so RTCP (Real-Time Control for managing and controlling them) Protocol) RTP data transfer and quality assurance depend on the lower layer protocol used.

  Therefore, in RTP, information such as synchronization lock information, data sequence number, and data type is also transferred. The user can reproduce the data in real time by having such information. The RTP application port number uses an even number, and RTCP uses an odd number one above. For this reason, voice and video conferences are connected by one network address and two port numbers.

  Even in data transfer using RTP, it is necessary to measure the presence or absence of packet loss in order to maintain a predetermined communication quality.

  FIG. 9 is a conceptual diagram showing an example of packet loss measurement in such a data communication system using RTP. In FIG. 9, the receiving terminal 2 receives a plurality of RTP packets transmitted from the transmitting terminal 1, and counts the number of receptions. The transmitting terminal 1 transmits the number of transmitted RTP packets on the RTCP packet transmission report TR periodically (for example, 100 msec), and the receiving terminal 2 receives it.

  The receiving terminal 2 calculates the packet loss number from the difference between the RTP packet reception number count and the transmission packet number obtained from the RTCP packet transmission report TR, and divides this by the expected reception number to calculate the loss ratio. A reply is periodically sent to the transmitting terminal 1 on the packet reception report RR. Similarly, the total number of lost packets after the start of reception is returned on the RTCP packet reception report RR.

FIG. 10 is a conceptual diagram showing another example of packet loss measurement in a data communication system using RTP. In FIG. 10, the transmitting terminal 1 transmits information on the number of scheduled transmission packets of an RTP packet to be transmitted from now on, and the receiving terminal 2 receives this and determines that the reception of the RTP packet is started.
Next, the transmission terminal 1 sequentially transmits RTP packets. The receiving terminal 2 sequentially receives RTP packets and stores and stores RTP sequence numbers. The transmitting terminal 1 transmits the transmission end information to the receiving terminal 2, and the receiving terminal 2 receives this and determines that the reception is complete.

  After the reception ends, the receiving terminal sorts the sequence number sequence in the arrival order as shown in FIG. Then, the number of packet losses in the case where one or more sequence numbers are missing and the number of consecutive losses in which sequence numbers are defined in advance are calculated as the number of bursty packet losses.

  However, the configuration of FIG. 9 is a general measurement method using RTCP, but does not support counting of the number of bursty packet losses in which packet loss occurs continuously. Moreover, since the RTCP packet used simultaneously with the RTP packet is used, it can be applied only to the RTP packet.

  On the other hand, according to the configuration of FIG. 10, since the packet loss number and the bursty packet loss number are calculated after sorting, it takes time to obtain the result. In particular, when the number of data increases, processing is performed by a sorting algorithm, so that an increase in CPU load used as a calculation terminal and a longer calculation time are expected. Further, CPU capacity is required to calculate the number of occurrences of bursty packet loss and the maximum / minimum / average of bursty packet loss.

  That is, the configuration of FIG. 10 is satisfactory because it can be processed with sufficient time to process the data once stored in batch, but is not preferable when the result is sequentially output in real time. .

  A problem to be solved by the present invention is a network quality evaluation measurement method that can be applied to loss measurement of all packets having sequence numbers, can perform load distribution, can perform real-time measurement of a large number of packets, and can output sequence numbers of lost packets To realize a network quality evaluation apparatus.

The invention described in claim 1 for achieving such a problem is as follows.
In measuring the communication quality of a network in which packets with consecutive sequence numbers flow,
Capture the packet flowing through the network, determine the difference in sequence number for the packets before and after the sequential, and detect the occurrence of packet loss based on the difference in real time ,
A network quality evaluation measurement characterized by deleting a corresponding number from a lost packet sequence number list stored in packet loss detection information when a packet with a sequence number determined to be a packet loss is delayed within a predetermined time. Is the method.

Invention of claim 2, wherein the perform presence detection of packet loss generates a trigger interval as the measurement interval, the network of claim 1, wherein performing a statistical processing of the number of packet losses that have occurred in each measurement cycle This is a quality evaluation measurement method.

Third aspect of the present invention, the number of packet losses in statistical processing, the cumulative number of packet losses, cumulative burst packet loss rate and these maximum values, minimum values, characterized in that it comprises at least one of the mean value A network quality evaluation measurement method according to claim 2 .

Fourth aspect of the present invention, and generates said performs presence detecting packet loss generating a trigger interval as the measurement interval, the detection information of the series of cuts every packet loss of packets in each measurement cycle The network quality evaluation measurement method according to any one of claims 1 to 3.

Invention of claim 5, wherein, the detection information of the packet loss, expiry set time, according to claim 4, characterized in that it comprises first number of loss packet sequence number, at least one of the loss packet sequence number list Network quality evaluation measurement method.

The invention described in claim 6
A network quality evaluation apparatus for measuring communication quality of a network through which packets with consecutive sequence numbers flow,
Means for sequentially capturing packets flowing through the network;
Means for determining the sequence number difference between the captured packets before and after,
Means for detecting in real time whether packet loss has occurred based on the difference;
A means for performing the statistical processing of the number of packet losses occurring at each measurement period, performing the trigger interval as a measurement period to detect the presence or absence of occurrence of packet loss,
A means for detecting the presence or absence of occurrence of packet loss with a trigger interval as a measurement period, and generating packet loss detection information for each break in a series of packet groups within each measurement period;
Means for storing generated packet loss detection information;
It is provided with.

  Advantageous Effects of Invention According to the present invention, a network quality evaluation measuring method and a network quality evaluation apparatus that can be applied to loss measurement of all packets having sequence numbers, can perform load distribution, can perform real-time measurement of a large number of packets, and can output sequence numbers of lost packets. Can be realized.

  Hereinafter, the present invention will be described in detail with reference to the drawings. FIG. 1 is a conceptual diagram of packet loss measurement according to the present invention. When the transmission terminal 1 transmits the transmission start information to the reception terminal 2, the reception terminal 2 enters a reception state. In this state, the transmission terminal 1 sequentially transmits packets with sequence numbers. The receiving terminal 2 performs real-time packet loss measurement according to the present invention based on the sequence number of the received packet. When the transmission terminal 1 finishes transmitting the packet with the sequence number, the transmission terminal 1 transmits “transmission end” information to the reception terminal 2. The receiving terminal 2 ends the measurement by receiving the “transmission end” information transmitted from the transmitting terminal 1.

  The receiving terminal 2 divides the period from the start to the end of transmission at several intervals, and calculates the number of packet losses and the number of bursty packet losses in real time as interval statistics for each interval.

  FIG. 2 is a block diagram illustrating a configuration example of a main part of the receiving terminal 2. Packets transmitted from the transmission terminal 1 are classified by data type by the filter unit 3 and a sequence number is extracted. The extracted sequence number is sequentially input to the shift register 4. The shift register 4 is provided with at least a current sequence number storage area 4a and a previous sequence number storage area 4b. The new sequence number is input to the current sequence number storage area 4a, and each time a new sequence number is input to the current sequence number storage area 4a, the sequence number that has been stored in the current sequence number storage area 4a is used as the previous sequence number. The data is sequentially transferred and stored in the previous sequence number storage area 4b.

The current sequence number and the previous sequence number stored in the shift register 4 are input to the sequence number difference calculation unit 5,
Δd = Sequence number difference Δd is calculated according to the equation: current sequence number−previous sequence number. The calculation result Δd of the sequence number difference calculation unit 5 is input to the memory control unit 6, and various data processing for obtaining packet loss information and bursty packet loss information, which will be described separately, is performed according to the size. .

  The memory control unit 6 includes a trigger generation unit 7 that outputs a trigger event signal at an arbitrary interval, and a bursty packet that sets how many consecutive packet losses are defined as bursty packet loss. A loss number setting unit 8 and a timeout setting unit 9 for setting a predetermined time for monitoring the expiration of the delayed packet are connected.

  The memory control unit 6 generates detection information in the memory 10 by detecting the detection result of the packet loss generated during the interval of the trigger event signal output from the trigger generation unit 7 for each break of a series of received packet groups. Store. For the multiple pieces of detection information generated during the interval of the trigger event signal, information on packet loss and bursty packet loss at each interval is obtained based on the setting values of the bursty packet loss number setting unit 8 and the timeout setting unit 9. Store. The memory control unit 6 also performs replenishment processing of lost packets using delayed packets.

  The detection information calculation unit 11 calculates the maximum value, minimum value, average value, etc. during each interval based on the packet loss and bursty packet loss information of each detection information stored in the memory 10, and the network quality Output as an evaluation measurement result.

  3 and 4 are flowcharts for explaining the flow of operations in the receiving terminal 2 of FIG. Before starting reception, the timeout setting unit 9 sets a detection timeout value (SP1), and the bursty packet loss number setting unit 8 sets a definition value (for example, 3) of bursty packet loss (SP2). When these are set, the timer is waiting for a trigger (SP3).

  When a trigger event signal is output from the trigger generation unit 7 in this state, the memory control unit 6 determines whether or not the trigger is the first trigger (SP4). If it is the first trigger, reception starts (SP5). If it is not the first trigger, the process proceeds to reception end processing (SP6).

  In the reception end process, the detection information calculation unit 11 calculates the accumulated packet loss value and the accumulated burst from the detection information for each break of a series of packet groups in the current interval stored in the memory 10 at that time. Packet loss values and their maximum, minimum, and average values are calculated (SP7). After each calculated value is confirmed as a measurement result for each interval (SP8), the results are stored in a predetermined file (SP9), and the number of triggers that the memory control unit 6 has planned It is determined whether or not the process should be terminated based on whether the (number of intervals) has been processed or whether a forced termination instruction has been instructed (SP10). If the state is to be terminated, all detection information remaining in the memory at this point is deleted and the series of processes is terminated. If not, the process returns to the reception process after step SP5.

  The reception process will be described. When starting the reception process, the memory control unit 6 resets the cumulative packet loss number and the cumulative burst packet loss number remaining in the memory 10 as the measurement result in the previous interval (SP11). After the reset is completed, a sequence number is acquired for the received packets of a series of packet groups (SP12) and sequentially stored in the shift register 4.

  The sequence number difference calculation unit 5 performs a calculation represented by Δd = current sequence number−previous sequence number based on the current sequence number and the previous sequence number sequentially stored in the shift register 4 to obtain a sequence number difference Δd ( SP13).

  The memory control unit 6 determines whether Δd> 1 (SP14), Δd = 1 (SP15), and Δd = 0 for the sequence number difference Δd calculated by the sequence number difference calculating unit 5 (SP14). SP16) and Δd <0 (SP17) are sequentially judged.

  First, if Δd> 1, it is determined that a packet loss has occurred, and the detection information is stored in the memory 10 (SP18). Here, the memory 10 stores the time-out setting time (5 sec in the example of FIG. 5), the leading number of the lost packet sequence number (for example, as shown in the detection information 1 of FIG. 89), a lost packet sequence number list (89, 90, 91, 92, 93, 94, 95, 96 in the example of FIG. 5) and the like are saved. After storing the detection information in the memory 10, the previous packet sequence number of the shift register 4 is updated (SP19), and the process returns to step SP12 again. If Δd> 1 is not satisfied, it is determined whether Δd = 1 (SP15).

  If Δd = 1, it is determined that there is no packet loss (SP20), and the previous packet sequence number in the shift register 4 is updated (SP19). If Δd = 1 is not satisfied, the process proceeds to determination of whether Δd = 0 (SP16).

  When Δd = 0, it is determined that the same packet is transmitted in duplicate, the process for the memory 10 is ignored, a duplicate packet detection process is performed separately (SP21), and the process returns to step SP12 again. When Δd = 0 is not satisfied, the process proceeds to determination of Δd <0 (SP17).

  When Δd <0, it is determined that a packet that has been processed as a packet loss has arrived with a delay, and corresponds to the loss packet sequence number list in the detection information stored in the memory 10 as shown below. Whether or not there is a packet having a sequence number to be searched is performed, and a loss packet supplement process is performed.

  First, a search is performed to delete detection information that has expired from a plurality of detection information stored in the memory 10 (SP22). At this time, the detection information calculation unit 11 calculates the packet loss count and burst packet loss count stored in the detection information to be deleted, the cumulative packet loss count, and the cumulative packet loss count in the current interval. The maximum, minimum, average, and cumulative packet loss counts for bursty packet loss counts are reflected on the previous network quality evaluation measurement results (SP23), and those results are updated to the latest in the current interval. Is output as a network quality evaluation measurement result.

  For example, if the detection information to be deleted is as shown in FIG. 6, the current packet loss count is 5, and the current bursty packet loss count is 3 (if the packet loss is defined as burst when 3 or more consecutive packet losses). Then, 5 is added to the total number of packet losses in the current interval, and 3 is added to the total number of bursty packet losses. Then, after adding these, the maximum value, the minimum value, and the average value of the number of bursty packet losses are recalculated, and then the detection information to be deleted is deleted from the memory 10 (SP24). After deletion, the process returns to step SP12 again.

  If there is no time-out detection information, the process proceeds to the next step to supplement the packet loss (SP25). That is, a search is performed as to which detection information includes a sequence number list in which the arrived packet is included. For example, when a target sequence number (for example, 90) is found as shown in FIG. Remove the sequence number (eg 90) from the list.

  For example, as shown in the detection information 1 of FIG. 8, when all the packets stored in the sequence number list of the detection information as packet loss arrive later in time and can be supplemented (SP26), the process goes to step SP24. Go ahead and delete detection information 1 itself.

  If the target packet is not in any of the sequence number lists stored in the memory 10, it is determined that it has already been deleted from the memory by the time-out process in step SP22, and the process returns to step SP12 again.

  According to such a configuration, unlike conventional batch processing, it is only necessary to perform arithmetic processing only when a packet loss occurs, so that the load can be distributed and long-term data is measured by real-time measurement or the like. However, there is no temporary high load due to arithmetic processing, and it does not affect other tasks.

  In the above embodiment, an example of an RTP packet has been described. However, the present invention is not limited to this, and the present invention can obtain the effect of load distribution with a similar algorithm for a similar packet having a sequence number.

  As described above, the network quality evaluation measuring method and the network quality evaluation apparatus that can be applied to loss measurement of all packets having a sequence number, can perform load distribution, can perform real-time measurement of a large number of packets, and can output the sequence number of lost packets. Can be realized.

It is a conceptual diagram of the packet loss measurement based on this invention. 3 is a block diagram illustrating a configuration example of a main part of a receiving terminal 2. FIG. 6 is a flowchart for explaining the flow of operations in the receiving terminal 2. 6 is a flowchart for explaining the flow of operations in the receiving terminal 2. 7 is an explanatory diagram of detection information stored in a memory of the receiving terminal 2. FIG. FIG. 6 is an explanatory diagram of detection information deleted from the memory of the receiving terminal 2 due to time out. 7 is an explanatory diagram of detection information supplemented by a delayed packet in the memory of the receiving terminal 2. FIG. FIG. 6 is an explanatory diagram of detection information that is deleted from the memory of the receiving terminal 2 when all delayed packets have arrived. It is a conceptual diagram which shows an example of the conventional packet loss measurement in the data communication system using RTP. It is a conceptual diagram which shows the other example of the conventional packet loss measurement in the data communication system using RTP. It is a conceptual diagram which shows the rearrangement in the conventional packet loss measurement.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Transmitting terminal 2 Receiving terminal 3 Filter part 4 Shift register 5 Sequence number difference calculating part 6 Memory control part 7 Trigger generating part 8 Burst-like packet loss number setting part 9 Timeout setting part 10 Memory 11 Detection information calculating part

Claims (6)

In measuring the communication quality of a network in which packets with consecutive sequence numbers flow,
Capture the packet flowing through the network, determine the difference in sequence number for the packets before and after the sequential, and detect the occurrence of packet loss based on the difference in real time ,
A network quality evaluation measurement characterized by deleting a corresponding number from a lost packet sequence number list stored in packet loss detection information when a packet with a sequence number determined to be a packet loss is delayed within a predetermined time. Method. 2. The network quality evaluation and measurement method according to claim 1, wherein detection of the occurrence of packet loss is performed using a trigger interval as a measurement period, and statistical processing is performed on the number of packet losses that occur at each measurement period. 3. The network quality evaluation according to claim 2, wherein the statistical processing of the packet loss number includes at least one of a cumulative packet loss number, a cumulative bursty packet loss number, and a maximum value, a minimum value, and an average value thereof. Measuring method. The detection of occurrence of packet loss is performed using a trigger interval as a measurement period, and packet loss detection information is generated for each break of a series of packet groups in each measurement period. The network quality evaluation measurement method according to any one of the above. 5. The network quality evaluation measurement method according to claim 4, wherein the packet loss detection information includes at least one of a timeout setting time, a head number of a lost packet sequence number, and a lost packet sequence number list. A network quality evaluation apparatus for measuring communication quality of a network through which packets with consecutive sequence numbers flow,
Means for sequentially capturing packets flowing through the network;
Means for determining the sequence number difference between the captured packets before and after,
Means for detecting in real time whether packet loss has occurred based on the difference;
A means for performing the statistical processing of the number of packet losses occurring at each measurement period, performing the trigger interval as a measurement period to detect the presence or absence of occurrence of packet loss,
A means for detecting the presence or absence of occurrence of packet loss with a trigger interval as a measurement period, and generating packet loss detection information for each break in a series of packet groups within each measurement period;
Means for storing generated packet loss detection information;
Means for deleting the corresponding number in the lost packet sequence number list stored in the packet loss detection information when the packet of the sequence number determined to be a packet loss is delayed within a predetermined time;
A network quality evaluation and measurement apparatus comprising:

Images