JP2006246118A - Method and apparatus for discriminating quality deterioration - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and an apparatus for discriminating whether a position causing the quality deterioration of a video distribution application or the like is a user side or a NW side on the basis of IP layer information obtained on a certain point on the receiving side. <P>SOLUTION: The apparatus comprises a function 51 for collecting specific packet information on a certain point of the receiving side from the transmitting side, a function 52 for estimating the transfer quality (a packet loss rate, receiving side delay time) of an IP quality layer and a packet loss generation position (on the transmitting side or receiving side) and a function for judging whether a position causing deterioration such as delay and packet loss is the transmitting side or the receiving side in each flow. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a technique related to network performance management in a packet switching network such as the Internet, and in particular, quality management technique and monitoring for narrowing down whether the cause of quality degradation is on the network side or the user side. It relates to technology.

In order to explain the conventional technology, the following definitions are given as a supplement.
[Quality degradation]
Generally, an event in which user satisfaction with a provided service decreases is called quality degradation. Here again, it is assumed that the service is provided in a target unit (session (or flow) unit) on the assumption that there is a target provided quality.

In particular,
-In the specified time length (several seconds to tens of minutes), (IP quality layer transfer quality) in the quality layer,
-Assumed communication protocol or application (TCP, UDP, FTP, HTTP, etc.)

-The quality measure (packet loss rate, transfer delay time, delay time fluctuation, etc.) specified by the quality item specified in advance is
• A condition that falls outside the specified tolerance is called quality degradation.
Note that the above quality measure or allowable range may vary depending on the communication protocol and application. Below, regardless of the communication protocol and application, the quality measure is the packet loss rate and transfer delay time, and the allowable range is specified uniformly. A description will be given assuming the case.

[Quality degradation isolation]
End-to-end quality targets (the upper and lower limits of the allowable range for the above quality scale) are allocated to each section, and the allowable range for each section is set. In addition, a section that cannot satisfy the target value allocated to the section is called a deterioration section, and a quality item that does not reach the target value is called a deterioration factor. That is, the quality deterioration isolation in the present invention is to narrow down the deterioration section and the deterioration factor.

Based on the above definitions, the following techniques can be cited as conventional techniques for identifying causes of quality degradation.
(1) Technology for acquiring MIB (Management Information Base) information (number of passed packets, amount of information, number of lost packets in buffer, etc.) implemented by a device or the like by SNMP (Simple Network Management Protocol) (for example, non-patent Reference 1).
(2) A technique for measuring performance (end-to-end packet loss rate, delay time, etc.) by installing a measuring device end-to-end of a network and performing test communication between them (for example, Non-Patent Document 2) reference).

The method (1) has a small effect on actual communication traffic because it is passive measurement, but cannot measure the MIB information of the equipment installed on the user side as a network operator, and the quality status of each user unit There is a problem that it is difficult to grasp and isolate deterioration.
In the method (2), there is a problem that a measuring device must be arranged at the end and end, and a test packet is applied to the network for measurement, which affects the actual communication traffic.

In addition to the above, there is a technique for transmitting a test packet from a certain point to an opposite point such as Ping and measuring the packet loss rate and the round-trip delay time in the round-trip route.
With this method, there is no need to place a measurement device at the end / end of the network, but because icmp packets are used, measurement accuracy in applications using TCP, only round trip delay can be measured, and a wide-area complex network is required. There is a problem that it is not suitable for the configuration.

Therefore, in the present invention, attention has been paid to calculating network quality information (packet loss, user-side delay time, etc.) from information passively measured at one point (on the receiving side) of the network and estimating the cause of deterioration.
By Steve Maxwell SNMP Network Management Tool Kosaido, 2001.p2-21 Takadoi, Kawamura, and Shinomiya: Evaluation of the effect of test packet transmission patterns on video streaming quality degradation detection, 2002, IEICE, pp.627

  The present invention solves the above-mentioned problems, and its purpose is to determine the cause of quality degradation of a video distribution application or the like based on IP layer information obtained at a certain point on the receiving side. It is to provide a method and an apparatus for discriminating whether it is on the NW side.

  In order to achieve the above object, the present invention collects specific packet information at a certain point from the transmission side to the reception side, transfers the IP quality layer (packet loss rate, reception side delay time), and generates packet loss. It has a function to estimate the location (transmission side or reception side) and a function to determine for each flow whether the cause of deterioration such as delay or packet loss is the transmission side or the reception side from the obtained quality information Features.

  Further, in the present invention, when performing quality degradation isolation, the degradation period is identified by the end-to-end permissible value of the packet transfer quality (packet loss rate) and the quality estimation measure (packet loss frequency, receiving side) in each section. It is characterized in that the deterioration section is specified based on the deterioration determination threshold value of (response time excess frequency).

  That is, the quality degradation isolation method according to the present invention first calculates end-to-end packet transfer quality (packet loss, delay, etc.) information in a packet switching network, and then uses the packet transfer quality information calculated above in advance. Compared with the determination threshold value set in (2), it is characterized by having a function of determining an end-to-end degradation point.

  More specifically, according to the quality degradation isolation method according to the present invention, when data is lost in the middle in order to ensure data transfer by the function of the protocol or the function of the application positioned above the communication protocol, the data degradation is performed again. In a communication to be transferred, a single point observation type quality degradation isolation method in a packet switching network that measures communication quality by monitoring at a certain point (observation point) between a transmission side and a reception side, and packet header information A flow extraction step that extracts the flow to be measured (from the start to the end of communication between the transmitting and receiving terminals), the location where packet loss and delay occur, and for each sequence section that divides the space consisting of sequence numbers Of the installed counters, operate the counter corresponding to the section to which the corresponding packet belongs to An NW quality estimation step for estimating the NW quality at the end, and a degradation location determination step for determining whether the degradation location is a reception side or a transmission side using the counter information after the operation as a base point. It is characterized (claim 1).

  In addition, the quality degradation isolating device in the packet switching network according to the present invention is configured so that the data is lost again when data is lost in the middle to ensure data transfer by the function of the protocol or the function of the application located above the communication protocol. In communication for transfer, a single point observation type quality degradation isolating device in a packet switching network for measuring communication quality by monitoring at a certain point (observation point) between a transmission side and a reception side, Flow extraction means that extracts the flow to be measured from the header information (from the start to the end of communication between the transmitting and receiving terminals), the sequence section that estimates the location where packet loss and delay occur, and divides the space consisting of sequence numbers The counter corresponding to the section to which the corresponding packet belongs is controlled. And NW quality estimation means for estimating end-to-end NW quality, and using the counter information after the operation, deterioration point determination means for determining whether the deterioration point is a reception side or a transmission side based on the observation point (Claim 2).

  Here, when the NW quality estimation means uses the TCP protocol, when determining the occurrence location of the packet loss, the ACK having the same sequence number is transmitted from the receiving side a number of times greater than or equal to the threshold value. Assume that a packet loss has occurred on the flow, search the range of the maximum window size in the retained data, determine whether there is a corresponding packet within this search range, and increase the counter value corresponding to the loss location (Claim 3).

  In addition, when the NW quality estimation unit uses the TCP protocol, the TCP transmission packet that first passes through the observation point is used as a base point when determining the occurrence of the delay on the user side. The time when the ACK packet to pass through the observation point is set as the ACK response time, and it is determined whether or not the ACK response time is equal to or greater than a preset time threshold value. (Claim 4).

  Further, the NW quality estimation means holds a determination threshold corresponding to a measuring device installation pattern such as NW-NW, NW-User, etc., and corresponding installation based on an average value of ACK response times during communication monitoring. It is preferable to have a mechanism that uses a pattern determination threshold (claim 5).

  In addition, the degradation point determination unit compares the counter information with a predetermined determination threshold, and determines whether a packet loss occurrence point of traffic in the reception direction to the user of the TCP protocol is on the transmission side with reference to the observation point. It is preferable to have a mechanism for determining whether it is on the receiving side (Claim 6).

  Further, the degradation point determination means may have a mechanism for comparing the counter information with a predetermined determination threshold and determining whether or not a delay occurs on the receiving side in TCP protocol transmission / reception traffic. Preferred (claim 7).

  Further, it is preferable that the deterioration location determination unit has a mechanism for simultaneously determining occurrence locations of a plurality of quality deterioration factors such as packet loss, delay, delay fluctuation, and the like for the same flow using the counter information. 8).

  When the RTP protocol is used as the protocol, the degradation point determination means uses the number of retransmitted packets and the number of normally arrived packets determined by the retransmit packet determination, and the packet loss of traffic to the receiving side is observed. It is preferable to have a mechanism for determining whether the point is on the upstream side or the downstream side with respect to the point (claim 9).

  Preferably, the NW quality estimation unit and the degradation point determination unit have a mechanism capable of determining the degradation point not only in the reception direction traffic to the user but also in the transmission direction traffic from the user. ).

According to the present invention, it is possible to provide a quality degradation isolation method and apparatus that do not require the installation of a measurement device for each ground of the network and that do not need to send out test packets or the like because of passive measurement.
By using this, the network operator can collect user-specific network information without depending on the user-owned device.

Further, from the standpoint of the network operator, it is possible to determine whether the user is caused by the own NW side or the user side, and this can be used as information for troubleshooting and complaint handling.
In addition, the network operator can acquire and manage information when the reception quality of the user is deteriorated and the deterioration classification is performed.

Hereinafter, the present invention will be described in detail based on preferred embodiments shown in the accompanying drawings.
First, a case where the TCP / IP protocol is used will be described.

FIG. 1 is a diagram showing a system configuration having a quality deterioration isolation function according to an embodiment of the present invention.
In the figure, 1 is a user (terminal), 2 is a communication partner (terminal), 3 is a network (User-NW) on the user side, 4 is a wide area network such as WAN, and 5 is a single point observation type according to the present embodiment. 1 shows a measuring instrument incorporating a quality degradation separation device (hereinafter also simply referred to as a separation device).

As shown in detail in FIG. 2, the separation device 5 according to the present embodiment has conventionally used a packet defined by a communication source address, a communication source port number, a communication destination address, a communication destination port number, and the like. A quality estimation function 52 and a degradation point determination function 53 as described later are added to the measuring instrument (measurement function 51) having the flow extraction function.
Hereinafter, each of the above functions will be described in detail.

First, the measurement function 51 will be described.
The setting of the measurement target in the measurement function 51 is specified by using, for example, the communication source address, the communication source port number, the communication destination address, the communication destination port number, the communication direction, and the like using the measurement target setting IF 54.

The measurement of the communication packet in the measurement function 51 can be realized by using Tcpdump, which is one of network packet capturing tools, for example. Information that can be acquired in this case is shown in FIG. Necessary information is an item marked with * at the beginning of the line in FIG.
Information shown in FIG. 3 (hereinafter referred to as packet information) is obtained every time a packet arrives. The sequence number is set or processed so as to be assigned from 1.
FIG. 4 shows an example of measurement information (output example of TCP / IP communication by Tcpdump).

Next, the flow extraction function will be described.
First, based on the parameters (communication source address, communication source port number, communication destination address, communication destination port number, communication direction) set as the measurement target as described above, the target arrival packet information is measured. Extract from The TCP protocol is identified here by looking at “protocol” in the IP header (see FIG. 5).

  Then, by looking at the TCP header information (see FIG. 3) of arrival packets in time series order, from the time of the packet whose flag is S (connection establishment time), the next flag is S (next connection establishment time) or The time until F (connection release time) and R (connection forced disconnection time) is defined as one flow.

Next, the quality estimation function 52 will be described.
The quality estimation function 52 is a function that divides the data into designated unit times based on the Tcpdump data acquired from the observation points described above, and calculates NW quality such as packet loss and delay for each unit time. is there. The quality estimation function 52 includes a packet loss estimation function (TCP / UDP) 52a and an ACK response time estimation function (TCP only) 52b (see FIG. 2), which will be described in order.

(1) TCP packet loss estimation function 52a:
In the implementation of the protocol, when it is determined that there is an unreceived packet on the receiving side, a plurality of ACK confirmation responses of unreceived sequence numbers are transmitted to the transmitting side (duplicate confirmation response). If this is used, it is possible to detect that a loss has occurred in the flow due to the fact that a duplicate confirmation response has been sent from the receiving side, and the missing sequence number is written in the duplicate confirmation response. Therefore, it is possible to identify missing packets using sequence numbers.

The operation flow of the packet loss estimation function 52a is shown in FIG. Based on the occurrence of a duplicate confirmation response, the received data area is searched retrospectively for the range of the sequence number for the window size, and the missing section depends on whether the transmission packet corresponding to the duplicate confirmation response has passed the observation point or not. It is determined whether the user side or the NW side.
In this search, if a missing packet is found (missing on the user side), the counter value A is incremented by +1, and if not found (missing on the NW side), the counter value B is incremented by +1. This search is repeated up to the total number N of ACKs during the measurement period.

(2) ACK response time estimation function 52b:
FIG. 7 shows an operation flow of the ACK response time estimation function 52b. In the TCP implementation, when the transmission packet reaches the reception side, the reception side returns the sequence number of the packet to be received next as an acknowledgment (ACK) to the transmission side. Therefore, the time when the transmission packet passes through the observation point and the positive acknowledgment corresponding to the transmission packet is returned from the reception side is defined as the ACK response time. Whether or not there is a delay can be detected.

However, since it is assumed that a transmission packet having the same sequence number arrives due to a retransmission packet, the response time is defined as the time when the first corresponding ACK packet arrives from the transmission packet that arrived at the observation point. Up to.
In addition, since the ACK response time in the steady state is used as a time threshold, a function for automatically correcting the time threshold for each user environment is required.
The response times of all ACK packets in the measurement interval are calculated. If the response time is longer than the time threshold, the counter A is incremented by +1, and otherwise, the counter B is incremented by +1.

FIG. 8 shows a specific example of the operation by the above-described TCP packet loss estimation function 52a and ACK response time estimation function 52b.
As shown in FIG. 8, when the packet # 1 is normally sent from the transmission side (upstream) to the reception side (downstream), the reception side (downstream) that has received the packet normally returns to the transmission side (upstream). An ACK packet (# 1 ACK packet) corresponding to the received packet is returned.
While this operation is repeated, packet communication continues normally.

  However, when the packet # 2 passes through the measurement point (observation point) and is lost for some reason on the downstream side, DupACK (duplicate response confirmation) sent from the reception side (downstream) to the transmission side (upstream) Packet) is detected at the observation point, and when the packet passes through the observation point, the packet loss on the downstream side is detected in this case.

  Similarly, after the # 3 packet passes through the observation point, the first response (ACK packet) sent from the reception side (downstream) to the transmission side (upstream) passes through the observation point. If the elapsed time exceeds the predetermined time, it is detected that there is a problem in the downstream NW although the ACK packet is transmitted.

Up to this point, the case where the TCP / IP protocol is used as the protocol has been described. Here, the case where the RTP protocol is used will be described.
(3) IP packet loss estimation function 52a in RTP packet transfer:
The IP packet loss situation (specifically, Kn and Rn below) is calculated by the following processes (A) to (C).

The calculation of output items is specified in advance with the sequence number as a unit. For example, C. This means that the IP packet loss situation is calculated every time the sequence number of the packet (s) sent by the transmission side increases by C.
Since the circulation problem due to the upper limit of the sequence number counter can be easily avoided by using the remainder class, here, the sequence number of the packet (data) transmitted first from the transmission side is set to 1.

(A) Determination of whether or not the packet is a retransmission packet:
RTP is a protocol on UDP, and unlike the TCP, the UDP layer does not perform recovery by retransmission when a packet is lost. However, if the AP has an ARQ (automatic retransmission control) function rather than UDP, recovery by retransmission is performed when a packet is lost. At that time, the same RTP sequence number is sent. The retransmission packet means an event of retransmitting an RTP sequence number sent in the past by the transmission source for recovery in the case of ARQ.

The following symbols are prepared for retransmission determination (see FIG. 9).
T ₀ : Measurement start time Pj: The jth packet arrived is called packet j tj: Pj arrival time Sj: Pj sequence number

m (t): Average increase rate of sequence number from measurement start T ₀ to time t (defined by increase of sequence number per unit time)
Smax (t): Maximum value among all sequence numbers of packets that arrived up to a certain time t Tmax (t): Time when the above packet arrived T: Retransmission determination parameter

The retransmission determination process will be described (see FIG. 10).
(Processing 1) Smax (t), Tmax (t), and m (t) are updated every time a packet arrives (time t).
Update of Smax (t) and Tmax (t) If the sequence number of the current arrival packet is larger than the maximum sequence number at the time of arrival of the previous packet, the value is updated.

-Update of m (t) For example, there are the following methods (a) to (d).
(A): Let m (t) be the slope of a straight line connecting the coordinates (T1, S1) consisting of the first packet arrival time and the sequence number and (Smax (t), Tmax (t)).
(B): An increase in the sequence number per unit time is calculated and set as m (t).
(C): In (b) above, the sequence number is limited to the most recent (for example, within ΔT) range.
(D): m (t) obtained by multiplying m (t) in (b) above by a safety factor.

(Processing 2) Judgment whether or not it is a retransmission packet (a) When it is an arrival packet within the first L (preset) th, retransmission judgment is not performed and all packets are considered not to be retransmission packets.

(B) For the Lth and subsequent arrival packets (assuming the jth arrival packet),

ならば再送パケットと判断する。
以上により再送か否かの判定ができる。

If so, it is determined as a retransmission packet.
As described above, it is possible to determine whether or not to retransmit.

(B) Count of Arrival Packets for Each Sequence Space The sequence number space is divided into designated constant C units, and sections A1, A2,... Are defined (referred to as sequence sections). For example, FIG. 11 shows an example when C = 10 ⁴ (bytes).

The network quality is calculated for each section An = [(n−1) · C, n · C]. This means that the quality index is calculated for each information unit (C) that the transmission side intends to transfer toward the reception side.
The processing here is to increase the following counter at the time of packet arrival.
Rn: counter of the number of retransmitted packets in section An Kn: counter of the number of non-retransmitted packets in section An

(Process 3)
For an arrival packet determined to be a retransmission packet, the sequence number is read, and the amount (in bytes) belonging to the corresponding section An is added to the counter Rn. However, when there are a plurality of sections, the following is performed on the counters for the plurality of sections.

However, # {·} is the number of sequence numbers included in the section {·}, and A∩B is a section common to the sections A and B.
For example, if section A = [5, 15] and section B = [10, 25],
# A = 10, # B = 15, A∩B = [10,15], # {A∩B} = 5

(Process 4)
For an arrival packet that has not been determined to be retransmitted, the sequence number is read, and the amount (in bytes) belonging to the corresponding section An is added to the counter Kn. However, when there are a plurality of sections, the following is performed on the counters for the plurality of sections.

但し、記号は処理3と同じである。

However, the symbols are the same as those in process 3.

(Process 5)
Using the counter values Kn and Rn at that time, the calculation is performed as follows.

但し、αは再送パケットが送信側に届くまでに、送信側が重複して何回送ったかを示すパラメータであり、例えば、高々１回の再送で届くならα＝１、通常はα＝１とする。
（注）正確には、上式の右辺は、「送信側における再送バイト数の割合（率）」であるが、ここでは、ＩＰ損失バイト数を上式で推定している。

However, α is a parameter indicating how many times the transmitting side has repeatedly sent the retransmission packet until it reaches the transmitting side. For example, α = 1 if it arrives at most once, and normally α = 1. .
(Note) To be exact, the right side of the above equation is the “ratio (rate) of the number of retransmission bytes on the transmission side”, but here, the number of IP loss bytes is estimated by the above equation.

(Process 6)
The section to which P _jmax belongs is _assumed to be Am = ((m−1) · C, m · C). Here, (m−1) · C <S _max (P _jmax ) <m · C.
For the unaggregated section of An (n <m), follow (Process 5).
Am is corrected as follows so that the denominator is not excessive.

但し、Ｃ’＝Ｓ_ｍａｘ（Ｐ_ｊｍａｘ）−（ｍ−１）・Ｃ
（注）上と同じ

However, C ′ = S _max (P _jmax ) − (m−1) · C
(Note) Same as above

(C) Calculation of network quality information Using the functions (A) and (B) described above, the IP packet loss situation is calculated as follows.
First, counters Kn, Rn (n = 1, 2,...) Are used for the arrival packet counters of sections A1, A2,.・ ・ Initialize N1) and prepare it.
Next, it is assumed that “the arrival of a packet belonging to the section An, n ≧ n2” is an event serving as an aggregation trigger of the counters K1, R1.

That is, although illustration is omitted, it is confirmed for each arrival packet whether the section to which the arrival packet belongs is An2 or later,
1) If it belongs to the section after section An2, the counters K1 and R1 are summed up, and the quality index in the section is executed as shown in process 5. After the processing, the counter situation is released, and new counters Kn1 + 1 and Rn + 1 are newly provided. This process realizes a mechanism for saving memory resources by dynamically adapting the counter. This mechanism is not affected by the transfer speed of the NW.

That is, since the trigger of the aggregation processing of a certain section is set as the time when the packet belonging to the section of N2 (parameter) destination arrives for the first time, unlike the case where the preset time (parameter) has passed, There is an advantage that it is not necessary to change the value of the setting parameter according to the number of arrival packets per unit time.
Further, in order to set an event that triggers the counting of the counters K2 and R2 to be processed next, it is assumed that n2 = n2 + 1, that is, n2 = N2 + 1.

2) If it belongs to a younger section, the above-described processes (A) and (B) are performed.
This process is repeated until the last arrival packet (j = Jmax).
Rn and Kn are obtained by the above procedure.

Next, the deterioration location determination function 53 will be described.
In this function, the degradation point for each end-to-end flow is determined using the counter value calculated by each function of the quality estimation function 52 described above.
Hereinafter, each situation will be described in the order of TCP and UDP.

(1) In the case of TCP protocol download direction (a) Packet loss estimation function 52a:
FIG. 12 shows a determination flow of the packet loss estimation function. FIG. 13 shows an explanatory diagram of the determination result.
After the search for all missing packets in the measurement section is completed, the determination is performed using the loss determination thresholds B1 and B2. Here, the determination threshold values B1 and B2 are fixed values because they do not depend on the user environment. N is the total number of ACKs in the measurement interval.

1) When (A + B) / N> B1, there is a high possibility that a loss frequently occurs in either (or both) NW or User NW within the measurement interval (determination (3)), and (A + B) / N < If it is B1, no loss has occurred (determination (4)).
2) When A (B) / N> B2, there is a high possibility that many losses occur at UserNW (NW) (determinations (1) and (2)).
When A / N> B2 and B / N> B2, it is highly possible that a loss has occurred in both UserNW and NW (determination (3)).

(B) ACK response time estimation function 52b:
FIG. 14 shows a determination flow of the ACK response time estimation function. FIG. 15 shows an explanatory diagram of the determination result.
After the measurement of all response times in the measurement section is completed, the determination is performed using the determination threshold C2. Since the judgment threshold C2 may depend on the installation environment, the installation pattern such as installation between NWs and NW-Users is determined from the response time during installation (steady state), and multiple determination thresholds are selected. It shall be possible.

At this time, if A / (A + B)> C2, a delay occurs in UserNW (determination (1)).
If A / (A + B) <C2, no delay has occurred in User NW (determination (2)).
(2) TCP protocol upload direction

(A) Packet loss estimation function 52a:
In the same way as above, even in the upload direction, B1 and B2 are used as determination thresholds.
1) When (A + B) / N> B1, there is a high possibility that there are many losses in either the NW or the User NW within the measurement interval.
If (A + B) / N <B1, no loss has occurred (determination (4)).

2) When A (B) / N> B2, there is a high possibility that many losses occur in UserNW (NW) (determinations (1) and (2)).
When A / N> B2 and B / N> B2, it is highly possible that a loss has occurred in both User NW and NW (determination (3)).

(B) ACK response time estimation function 52b:
Similarly, in the upload direction, C2 is set as a determination threshold value.
If A / (A + B)> C2, there is a high possibility that a delay has occurred in the NW (determination (1)). If A / (A + B) <C2, there is no delay in NW (determination (2)).

(3) Download direction of RTP packet transfer The determination of the quality degradation location in the section to be evaluated is made as follows.
D1, D2, and D3 are set as determination thresholds.
(A) When Rn = 0 (that is, when there is no ARQ function or there is an ARQ function and packet loss does not occur anywhere end-to-end)
The following situations are detected and carved out.

(A1) When 1−Kn / C <D1 (that is, many packets that are not retransmitted packets pass through almost all measurement points), there is a high possibility that no packet loss has occurred in the NW. judge.
(A2) If Kn / C <D2 (that is, there is little passage of packets that are not retransmission packets), it is determined that a packet loss has occurred in the NW.
(A3) Other than above Judgment hold

(B) When Rn> 0 (that is, when there is an ARQ function and packet loss occurs somewhere end-to-end)
The following situations are detected and carved out.
(B1) No deterioration when Rn / C ≦ D3 (b2) When 1−Kn / C <D1 and Rn / C> D3 (that is, many packets that are not retransmitted packets pass through almost the measurement point) If it is determined that the packet loss is likely to occur in the User NW.

(B3) When Kn / C <D2 and Rn / C> D3 (that is, there is little passage of packets that are not retransmission packets), it is determined that there is a high possibility that a packet loss has occurred in the NW.
(B4) In cases other than the above

(4) Upload direction of RTP packet transfer
(A) When Rn = 0 (that is, when there is no ARQ function, or when there is an ARQ function and packet loss does not occur anywhere end-to-end), the following situation is detected and isolation is performed.

(A1) When 1−Kn / C <D1 (that is, many packets that are not retransmitted packets pass through almost all measurement points), there is a high possibility that a user NW has a packet loss. judge.
(A2) If Kn / C <D2 (that is, there is little passage of packets that are not retransmission packets), it is determined that a packet loss has occurred in the NW.
(A3) Other than above Judgment hold

(B) When Rn> 0 (that is, when there is an ARQ function and packet loss occurs somewhere end-to-end)
The following situations are detected and carved out.
(B1) No deterioration when Rn / C ≦ D3 (b2) 1−Kn / C <D1 and Rn / C> D3 (that is, many packets that are not retransmitted packets pass through almost the measurement point) If it is determined that a packet loss has occurred in the User NW.

(B3) When Kn / C <D2 and Rn / C> D3 (that is, there is little passage of packets that are not retransmission packets), it is determined that a packet loss has occurred in the NW.
(B4) In cases other than the above

  Each of the above-described embodiments shows an example of the present invention, and the present invention is not limited to these, and appropriate modifications and improvements may be made without departing from the spirit of the present invention. Needless to say.

It is a figure which shows the system configuration | structure which has the quality degradation isolation | separation function which concerns on one Embodiment of this invention. It is a block diagram which shows the detail of the carving apparatus which concerns on embodiment. It is a figure which shows the information which can be acquired in the measurement of the communication packet in the measurement function of the dividing device which concerns on embodiment. It is a figure which shows the specific measurement information example (output example of TCP / IP communication by Tcpdump) in a measurement function. It is a figure which shows the header structure of IPv4. It is an operation | movement flowchart of the packet loss estimation function in the quality estimation function which concerns on embodiment. It is an operation | movement flowchart of the ACK response time estimation function in the quality estimation function which concerns on embodiment. It is a figure which shows the specific example of the operation | movement by the TCP packet loss estimation function and ACK response time estimation function which concern on embodiment. It is explanatory drawing of the operation | movement at the time of the packet arrival in RTP packet transfer. FIG. 6 is a flowchart of a retransmission determination operation. It is explanatory drawing of an arrival packet count similarly. It is an operation | movement flowchart of the packet loss estimation function in the degradation location determination function which concerns on embodiment. It is explanatory drawing of a determination result similarly. It is an operation | movement flowchart of the ACK response time estimation function in the degradation location determination function which concerns on embodiment. It is explanatory drawing of a determination result similarly.

Explanation of symbols

1 user (terminal)
2 Communication partner (terminal)
3 User side network (User-NW)
4 Wide area network (WAN)
5 (one-point observation type) quality degradation isolating device 51 measurement function 52 quality estimation function 52a packet loss estimation function 52b ACK response time estimation function 53 degradation point determination function 54 measurement target setting IF

Claims (10)

There is a communication between the sending side and the receiving side in the communication that performs data transfer again when data is lost in the middle to ensure data transfer by the function of the protocol or the application function that is positioned above the communication protocol. A quality degradation isolation method in a packet switching network that measures communication quality by monitoring at one point (observation point),
A flow extraction step of extracting a flow to be measured from packet header information (from the start to the end of communication between transmission and reception terminals);
End-to-end NW quality is estimated by operating the counter corresponding to the section to which the corresponding packet belongs, among the counters installed for each sequence section that divides the space consisting of sequence numbers by estimating the location where packet loss and delay occur NW quality estimation step for estimating
A one-point observation type quality degradation isolation method, comprising: a degradation location determination step of determining whether a degradation location is a reception side or a transmission side using the observation point as a base point using the counter information after the operation. There is a communication between the sending side and the receiving side in the communication that performs data transfer again when data is lost in the middle to ensure data transfer by the function of the protocol or the application function that is positioned above the communication protocol. A quality degradation isolation device in a packet switching network for measuring communication quality by monitoring at one point (observation point),
A flow extraction means for extracting a flow to be measured from packet header information (from the start to the end of communication between transmission and reception terminals);
End-to-end NW quality is estimated by operating the counter corresponding to the section to which the corresponding packet belongs, among the counters installed for each sequence section that divides the space consisting of sequence numbers by estimating the location where packet loss and delay occur NW quality estimation means for estimating
A single-point observation type quality deterioration isolation device, comprising: deterioration point determination means for determining whether a deterioration point is a reception side or a transmission side using the observation point as a base point using the counter information after the operation. The NW quality estimation means is
When the TCP protocol is used, a packet loss has occurred on the flow because an ACK having the same sequence number has been sent more than a threshold number from the receiving side when determining where the packet loss has occurred. age,
The range of the maximum window size in the retained data is searched, the presence / absence of a corresponding packet in the search range is determined, and the counter value corresponding to the loss location is increased. A single point observation type quality degradation isolation device. The NW quality estimation means is
When using the TCP protocol, when determining the occurrence of the delay on the user side, the TCP transmission packet that first passed through the observation point as a base point,
The time when the ACK packet corresponding to the transmission packet first passes through the observation point is set as the ACK response time, and it is determined whether or not the ACK response time is equal to or greater than a preset time threshold. The single-point observation type quality deterioration isolating device according to claim 2, wherein the single point observation type quality deterioration isolating device has a mechanism for increasing the frequency. The NW quality estimation means is
A mechanism that holds determination thresholds according to measuring instrument installation patterns such as between NW-NW and between NW-User, and uses the determination threshold of the corresponding installation pattern based on the average value of ACK response times during communication monitoring. The single-point observation type quality deterioration isolating device according to claim 4. The deterioration point determination means is
A mechanism for comparing the counter information with a preset determination threshold and determining whether a packet loss occurrence location of traffic in the reception direction to a TCP protocol user is on the transmission side or the reception side based on the observation point The single-point observation type quality deterioration classification device according to claim 3, wherein: The deterioration point determination means is
5. The system according to claim 4, wherein the counter information is compared with a preset determination threshold value to determine whether or not a delay occurs on the receiving side in TCP protocol transmission / reception traffic. Single point observation type quality degradation isolation device. The deterioration point determination means is
5. The one-point observation type according to claim 3, wherein the counter information is used to simultaneously determine occurrence locations of a plurality of quality degradation factors such as packet loss, delay, delay fluctuation, and the like for the same flow. Quality degradation isolation device. When using the RTP protocol:
The deterioration point determination means is
Using the number of retransmitted packets determined by retransmission packet determination and the number of normally arriving packets, it has a mechanism to determine whether the packet loss of traffic to the receiving side is upstream or downstream with respect to the observation point The single-point observation type quality deterioration isolating device according to claim 2. The NW quality estimation means and the deterioration point determination means are
The single-point observation type quality deterioration according to any one of claims 2 to 9, characterized in that a deterioration location can be determined not only in the reception direction traffic to the user but also in the transmission direction traffic from the user. Carving device.

Images