JP4343792B2 - Network congestion scale determination method and system - Google Patents

Description

  The present invention consolidates packet repair states in FEC (Forward Error Correction) proxies (decoders) installed between end hosts in an area unit in bi-directional real-time communication between end hosts performed over an IP network. The present invention relates to a network congestion scale determination method and system for grasping the scale.

  With the recent spread of the Internet, demand for interactive real-time applications such as VoIP and video conferencing is increasing. Current Internet users use services by connecting various terminals to a network environment. In addition, many users use so-called best-effort services that have no quality guarantee regarding communication bandwidth, communication delay, etc., so when packets from multiple users concentrate on a certain point in the network. As a result, quality degradation such as packet loss occurs, which also affects the quality of the application. Therefore, studies for maintaining quality by applying a code error correction technique (FEC) to bi-directional real-time applications are currently in progress (see Non-Patent Document 1, for example).

Shigeru Teruhi and two others “Applicability of FEC to interactive real-time applications”, 2003 IEICE Communication Society, B-7-33

  However, FEC that corrects lost packets during communication via the IP network from the received packets on the receiving side only performs the packet loss repair process to stabilize the quality for each session. However, it does not avoid the congestion due to the over-accommodation of the network, which is the cause of packet loss, and there is a limit to the stabilization of quality only by FEC. For this reason, it is necessary to implement other quality control methods such as accommodation restriction and detour control related to the network while performing quality stabilization by FEC.

  The present invention has been made to solve the above-mentioned problems, and its purpose is to determine the quality of each session passing through the same congestion point in order to serve as a trigger for another quality control method, and its end host. The congestion or normal state is determined for each area that accommodates the data, and the congestion scale is grasped when there is further congestion.

In order to achieve the above object, the invention according to claim 1 is a session between an end host and an FEC proxy including an encoder and a decoder for each end host to perform error correction processing corresponding to the end host, respectively. Connected to an IP network via a quality measurement means for measuring quality, and connected to the network congestion status that occurs when bidirectional real-time communication is performed between multiple end hosts via FEC proxies. A network congestion scale determination method for measuring by a congestion scale determination means, wherein the method measures a packet repair state at a decoder in an FEC proxy corresponding to the quality measurement means by each of the quality measurement means, and a session Determining the quality state of the unit, and determining the unit determined by each of the quality measuring means. The quality status of Deployment unit aggregated by the congestion scale determining means determines the congestion state of each area of the network having the same convergence point, a predetermined value following scale congestion state of the discriminated by a predetermined period monitoring state A network congestion scale determination method comprising: determining a recovery state when maintaining
Determining the session-by-session quality state comprises:
Measure the number of repaired packets where the FEC proxy decoder repaired lost packets by code error correction processing, or the number of unrepairable packets that could not be repaired by code error correction processing, and defined at least one threshold value. Determining a plurality of predictive levels by comparing the measured quality state of the session and the threshold, and dividing the session state into a plurality of quality levels,
The step of determining the congestion state or the recovery state of the area unit,
The quality level in units of sessions determined by each quality measuring unit is managed in units of areas by associating each end host with an area, and at least one threshold for the session ratio for each quality level in the area And a value representing the largest congestion size among a plurality of predetermined congestion sizes is determined from a combination of a plurality of predictive levels and a plurality of session ratios, and the determined congestion size is determined for the network. And determining a congestion state or a recovery state of the area.

Further, the invention according to claim 2 is the network congestion scale determination method according to claim 1 , wherein the step of determining the recovery state from the congestion state of each area unit is configured to cause the congestion scale determination means to determine the congestion scale for a certain period. The method includes a step of setting a protection time for monitoring, and determining a normal state when a state in which the session ratio is always equal to or lower than a predetermined threshold is maintained throughout the protection time.

The invention according to claim 3 is the network congestion scale determination method according to claim 1 or 2 , wherein each of the quality measuring means and the FEC proxy is provided on a network path in each end host or in a user premises of each end host. It is provided in either of these.

Further, the invention of claim 4 is the quality measurement for measuring the session quality between the end host and the FEC proxy including the encoder and decoder for each end host to perform error correction processing corresponding to the end host, respectively. The network congestion status that occurs when two-way real-time communication is performed between multiple end hosts via the FEC proxy is measured by the congestion scale determination means connected to the network. Network congestion scale determination system, wherein each of the quality measurement means has a function of measuring a packet repair state at a decoder in an FEC proxy corresponding to the quality measurement means and determining a quality state in units of sessions. , The congestion scale determining means determines the session unit determined by each quality measuring means. A network congestion scale determination system having a function of determining the congestion state or the recovery state of the area units of the network that aggregates the quality state, and has the same convergence point,
Each of the quality measuring means is
Measure the number of repaired packets where the FEC proxy decoder repaired lost packets by code error correction processing, or the number of unrepairable packets that could not be repaired by code error correction processing, and defined at least one threshold value. A plurality of predictive levels are determined by comparing the measured session quality state with the threshold, and the session state is divided into a plurality of quality levels,
The congestion scale determining means is
The quality level in units of sessions determined by each quality measuring unit is managed in units of areas by associating each end host with an area, and at least one threshold for the session ratio for each quality level in the area And a value representing the largest congestion size among a plurality of predetermined congestion sizes is determined from a combination of a plurality of predictive levels and a plurality of session ratios, and the determined congestion size is determined for the network. A network congestion scale characterized by having a means for determining as a congestion state of an area, and determining a recovery state when the determined congestion state is monitored for a certain period of time and maintaining a state of a predetermined value or less. In the judgment system.

The invention according to claim 5 is the network congestion scale determination system according to claim 4 , wherein the congestion scale determination means has a function of determining a recovery state from a congestion state in an area unit in the congestion scale determination means. A protection time for monitoring the congestion level of time is set, and a function to determine a normal state when the session ratio is constantly kept below a predetermined threshold throughout the protection time is provided. It is characterized by.

The invention according to claim 6 is the network congestion scale determination system according to claim 4 or 5 , wherein the quality measuring means and the FEC proxy are provided on the network path in each end host or in the user premises of each end host. It is provided in either.

Further, the invention of claim 7 is the quality measurement for measuring the session quality between the end host and the FEC proxy including the encoder and decoder for each end host to perform error correction processing corresponding to the end host, respectively. The network congestion status that occurs when two-way real-time communication is performed between multiple end hosts via the FEC proxy is measured by the congestion scale determination means connected to the network. A program for causing a computer to execute a determination process of the network congestion scale to be performed by measuring a packet restoration state at a decoder in an FEC proxy corresponding to the quality measuring means by each quality measuring means, A step of determining a state, and the determination by each of the quality measuring means And the quality state of the session unit aggregated by the congestion scale determining unit, when executing the steps of: determining a congestion state or the recovery state of the area units of the network having the same convergence point, to the computer,
Measure the number of repaired packets where the FEC proxy decoder repaired lost packets by code error correction processing, or the number of unrepairable packets that could not be repaired by code error correction processing, and defined at least one threshold value. Determining a quality state for each session, comprising: determining a plurality of predictive levels by comparing the measured quality state of the session with the threshold; and classifying the session state into a plurality of quality levels; ,
The quality level in units of sessions determined by each quality measuring unit is managed in units of areas by associating each end host with an area, and at least one threshold for the session ratio for each quality level in the area And a value representing the largest congestion size among a plurality of predetermined congestion sizes is determined from a combination of a plurality of predictive levels and a plurality of session ratios, and the determined congestion size is determined for the network. When determining as the congestion state of the area unit including the step of determining as the congestion state or the recovery state of the area, and monitoring the determined congestion state for a certain period of time and maintaining a state of a predetermined value or less Determining a recovery state;
Is in a program that causes a computer to execute.

The invention according to claim 8 resides in a recording medium on which the program according to claim 7 is recorded.

  According to the present invention, it is possible to determine the quality of each session that passes through the same congestion point, determine the congestion or normal state for each area that accommodates the end host, and further understand the congestion scale at the time of congestion. In addition, by combining with other quality control methods such as accommodation restriction and bypass control, while performing repair processing by FEC, the congestion state is detected before the client quality is affected, and other control Network quality can be stabilized by using a trigger.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a schematic configuration diagram of an embodiment of a network congestion scale determination system according to the present invention. In this network congestion scale determination system, the end hosts 1a to 1j correspond to the network area 4a in the IP network 4c via the corresponding FEC proxies 2a to 2j and the quality measurement means 3a to 3j, respectively, and the end hosts 1k to 1t correspond to the end hosts 1k to 1t, respectively. The congestion scale determination means 5 is connected to the network 4c via the FEC proxies 2k to 2t and the quality measurement means 3k to 3t, to the network area 4b. In this example, end hosts 1a and 1k, 1b and 1l,..., 1j and 1t are connected via networks 4a, 4c and 4b, respectively, to implement interactive real-time applications such as video conferencing. Shall.

  FIG. 2 is limited to the implementation of the bidirectional real-time application between the end host 1a and the end host 1k shown in FIG. 1, and the functions of the FEC proxy 2a; 2k and the quality measuring means 3a; 3k are described in detail. FIG. Each FEC proxy 2a; 2k is an encoder that adds an error correction code to the data transmitted from the end host 1a; 1k corresponding to the FEC proxy by FEC encoding processing and forwards it to the end host 1k; 1a of the communication partner 6a; 6k, and an end host 1k of the communication partner; an FEC proxy 2k corresponding to the end host from 1a; an encoder 6k of 2a; a decoder 7a; 7k for performing FEC restoration processing of data transferred via 6a, respectively I have.

  The data transmitted from the end host 1a is added with an error correction code by the encoder 6a in the FEC proxy 2a and transmitted to the decoder 7k in the FEC proxy 2k. The decoder 7k that has received the data to which the error correction code is added performs a repair process according to the data loss and transmits the data to the end host 1k. Conversely, the data transmitted from the end host 1k is added with an error correction code by the encoder 6k in the FEC proxy 2k and transmitted to the decoder 7a in the FEC proxy 2a. The decoder 7a that has received the data to which the error correction code is added performs a repair process according to the data loss and transmits the data to the end host 1a.

  The quality measuring means 3a measures the number of repair packets and the number of unrepairable packets at the decoder 7a, and the quality measurement means 3k at the decoder 7k. The number of repair packets is the number of packets generated between the encoder 6a in the FEC proxy 2a and the decoder 7k in the FEC proxy 2k, or between the encoder 6k in the FEC proxy 2k and the decoder 7a in the FEC proxy 2a. This refers to the number of packets in which loss can be repaired by error correction processing in the decoder 7k or 7a and the packet received at the end host 1k or 1a has no loss. On the other hand, the number of unrepairable packets is received at the end host without being completely repaired by the error correction processing in the decoder because many packet losses occur between the encoder of one FEC proxy and the decoder of the other FEC proxy. This refers to the number of packets that will be lost.

  Next, each quality measuring means 3a and 3k compares the measured number of repaired packets or the number of unrepairable packets with at least one or more arbitrarily defined thresholds, thereby changing the session quality state between normal and congestion. It is determined that the state is classified into signs (levels 1 to n) according to the scale. In this embodiment, by setting three thresholds, normal (threshold 1 or lower), predictor (level 1: threshold 1 or higher, threshold 2 or lower), predictor (level 2: threshold 2 or higher, threshold 3 or lower), Assume that a four-stage determination of a sign (level 3: threshold 3 or higher) is made. Then, the determination result (normal / prediction (levels 1 to 3)) is notified to the congestion scale determination means 5. Of course, the other quality measuring means 3b to 3t and FEC proxies 2b to 2t in FIG. 1 have the same functions as the functions of the quality measuring means 3a; 3k and the FEC proxy 2a; 2k. It is.

  In this way, the congestion scale determination means 5 that receives the quality determination results (normal / predictors (levels 1 to 3)) from the quality measurement means 3a to 3t corresponding to the end hosts 1a to 1t Aggregate quality judgment results and judge congestion scale. As shown in FIG. 3, the congestion scale determination means 5 holds correspondence information between the network areas 4a and 4b and the quality measurement means 3a to 3t as notification sources of quality determination results in advance, and each quality measurement means 3a The quality measurement results (normal / predictors (levels 1 to 3)) from ~ 3t are managed in correspondence with this information. The quality determination result shown in FIG. 3 shows an example of the quality state of the session when bidirectional real-time communication is performed between corresponding end hosts in the network configuration of FIG.

  The communication from the end host 1a to the end host 1k in FIG. 2 is associated with the network area 4b as the determination result of the quality measuring means 3k because the repair process is performed by the decoder 7k. On the contrary, the communication from the end host 1k to the end host 1a is associated with the network area 4a as the determination result of the quality measuring means 3a because the repair process is performed by the decoder 7a. Based on this management information, the congestion scale determination means 5 periodically, for each session in the network area 4a, 4b, normal, predictor (level 1), predictor (level 2) and predictor (level 3) To calculate the session ratio. The ratio of each session in the network area 4a in FIG. 3 is calculated as 50% normal session, 50% sign (level 1), and 20% sign (level 2). Here, there are three precursor (level 1) sessions, but since the precursor (level 2) session is also in the state of the precursor (level 1) or more, the number of sessions is added and the ratio is calculated as five.

  Next, the congestion scale determination means 5 that has calculated the session ratio of each predictive level for each network area has at least one or more arbitrarily defined threshold values (for example, threshold 1: session ratio is the session ratio). 10% or less, threshold 2: session ratio is 20% or less, and threshold 3: session ratio is 30% or less. FIG. 4 shows an example of the congestion scale determination, which is classified into nine stages (1) to (9) according to the combination of the session quality level and the session ratio for each quality level. It is divided into three stages (congestion scale 1: (1) to (4) (7), congestion scale 2: (5) (6) (8), and congestion scale 3: (9)).

  When the session ratio for each quality level in FIG. 3 is assigned to FIG. 4, the sign (level 1) is 50% (3), and the sign (level 2) is 20% (5). Congestion scale 1 and congestion scale 2 are applied, and the larger one is applied, and network area 4a is determined to be congestion scale 2.

  On the other hand, the recovery from the congestion state is determined by setting a protection time for monitoring the congestion scale for a certain period in the congestion scale determination means 5. In FIG. 4, the state where the session ratio for each quality level is 10% or less is monitored for a certain period, and when the state is constantly maintained at 10% or less, it is determined as a normal state. With the above processing, the congestion scale determination means 5 determines the congestion scale, and the determination result can be used for other quality control such as accommodation restriction and detour control related to the network.

  Note that the FEC proxies 2a to 2t and the quality measuring means 3a to 3t corresponding to the end hosts 1a to 1t in the network congestion scale determination system described above are independently arranged in the user's home or the corresponding end hosts 1a to 1t. You may make it the structure provided for 1t.

  In addition, the end hosts 1a to 1t, the FEC proxies 2a to 2t, the quality measuring means 3a to 3t, and the congestion scale judging means 5 of the network congestion scale judging system shown in FIG. 1 and FIG. Each can be configured by a computer incorporating software. These software programs are stored in a storage medium such as a magnetic disk (floppy disk (registered trademark), hard disk, etc.), optical disk (CD-ROM, DVD, etc.), semiconductor memory, etc. as programs that can be executed by a computer. You can also

It is a schematic block diagram which shows an example of the network congestion scale determination system by this invention. It is a figure for demonstrating in detail the function of a FEC proxy and a quality measurement means. It is a figure which shows the example of the session management table which a congestion scale determination means manages. It is a figure which shows the example of congestion scale determination.

Explanation of symbols

1a ~ 1t End host
2a ~ 2t FEC proxy
3a to 3t Quality measurement means
4a-4c IP network 5 Congestion scale judgment means
6a, 6k FEC encoder
7a, 7k FEC decoder

Claims (8)

Each end host is connected to the IP network via an FEC proxy including an encoder and a decoder for performing error correction processing corresponding to the end host and a quality measuring means for measuring session quality between end hosts. A network congestion scale determination method for measuring a congestion state of a network generated when bidirectional real-time communication is performed between a plurality of end hosts via an FEC proxy by a congestion scale determination unit connected to the network. The method measures the packet repair state at the decoder in the FEC proxy corresponding to the quality measuring means by the quality measuring means, and determines the quality state in units of sessions, and the quality measuring means determines The quality level of each session is determined by the congestion scale judging means Promises, a network congestion scale determination method having a step of determining a congestion state or the recovery state of the area units of the network having the same convergence point,
Determining the session-by-session quality state comprises:
Measure the number of repaired packets where the FEC proxy decoder repaired lost packets by code error correction processing, or the number of unrepairable packets that could not be repaired by code error correction processing, and defined at least one threshold value. Te, comprising the steps of defining a plurality of sign levels by comparing the quality state of the session that the measurement threshold, dividing the state of the session to a plurality of quality levels,
The step of determining the congestion state or the recovery state of the area unit,
The quality level in units of sessions determined by each quality measuring unit is managed in units of areas by associating each end host with an area, and at least one threshold for the session ratio for each quality level in the area And a value representing the largest congestion size among a plurality of predetermined congestion sizes is determined from a combination of a plurality of predictive levels and a plurality of session ratios, and the determined congestion size is determined for the network. Determining a congestion state of the area, and monitoring the determined congestion state for a certain period and determining a recovery state when maintaining a state of a predetermined value or less. Congestion scale judgment method.   The step of determining a recovery state from the congestion state of the area unit sets a protection time for monitoring the congestion scale for a certain period in the congestion scale determination means, and the session ratio is always predetermined during the protection time. The network congestion scale determination method according to claim 1, further comprising a step of determining a normal state when a state equal to or lower than a threshold value is maintained.   The network congestion scale determination method according to claim 1 or 2, wherein each of the quality measurement means and the FEC proxy is provided either on each end host or on a network path in a user residence of each end host. Each end host is connected to the IP network via an FEC proxy including an encoder and a decoder for performing error correction processing corresponding to the end host and a quality measuring means for measuring session quality between end hosts. A network congestion scale determination system that measures a congestion state of a network that occurs when two-way real-time communication is performed between a plurality of end hosts via an FEC proxy using a congestion scale determination unit connected to the network. Each of the quality measuring means has a function of measuring a packet repair state at a decoder in the FEC proxy corresponding to the quality measuring means and determining a quality state in units of sessions, and the congestion scale determining means includes the Aggregate the quality status of each session determined by each quality measurement means, and A network congestion scale determination system having a function of determining the congestion state or the recovery state of the area units of the network having the same convergence point,
Each of the quality measuring means is
Measure the number of repaired packets where the FEC proxy decoder repaired lost packets by code error correction processing, or the number of unrepairable packets that could not be repaired by code error correction processing, and defined at least one threshold value. Te, defining a plurality of sign levels by comparing the quality state of the session that the measurement threshold, and means for dividing the state of the session to a plurality of quality levels,
The congestion scale determining means is
The quality level in units of sessions determined by each quality measuring unit is managed in units of areas by associating each end host with an area, and at least one threshold for the session ratio for each quality level in the area And a value representing the largest congestion size among a plurality of predetermined congestion sizes is determined from a combination of a plurality of predictive levels and a plurality of session ratios, and the determined congestion size is determined for the network. A network congestion scale characterized by having a means for determining as a congestion state of an area, and determining a recovery state when the determined congestion state is monitored for a certain period of time and maintaining a state of a predetermined value or less. Judgment system.   The function of determining the recovery state from the congestion state in units of areas in the congestion scale determination unit sets a protection time for monitoring the congestion scale for a certain time in the congestion scale determination unit, and the session is performed during the protection time. 5. The network congestion scale determination system according to claim 4, wherein the network congestion scale determination system has a function of determining a normal state when the ratio is constantly maintained at a predetermined threshold value or less.   6. The network congestion scale determination system according to claim 4, wherein the quality measuring means and the FEC proxy are provided in each end host or on a network path in a user residence of each end host. Each end host is connected to the IP network via an FEC proxy including an encoder and a decoder for performing error correction processing corresponding to the end host and a quality measuring means for measuring session quality between end hosts. Network congestion scale determination processing for measuring the congestion state of a network generated when two-way real-time communication is performed between a plurality of end hosts via an FEC proxy by a congestion scale determination means connected to the network. Measuring the packet repair state at the decoder in the FEC proxy corresponding to the quality measurement means by each quality measurement means, and determining the quality state in units of sessions, Quality per session determined by the quality measure The status aggregation by the congestion scale determining unit, when executing the step of determining a congestion state or the recovery state of the area units of the network having the same convergence point, to the computer,
Measure the number of repaired packets where the FEC proxy decoder repaired lost packets by code error correction processing, or the number of unrepairable packets that could not be repaired by code error correction processing, and defined at least one threshold value. Te, defining a plurality of sign levels by comparing the quality state of the session that the measurement threshold, comprising the steps of dividing the state of the session to a plurality of quality levels, determining the quality state of the session ,
The quality level in units of sessions determined by each quality measuring unit is managed in units of areas by associating each end host with an area, and at least one threshold for the session ratio for each quality level in the area And a value representing the largest congestion size among a plurality of predetermined congestion sizes is determined from a combination of a plurality of predictive levels and a plurality of session ratios, and the determined congestion size is determined for the network. Determining the congestion state of the area, and determining the recovery state when the determined congestion state is monitored for a certain period of time and maintaining a state of a predetermined value or less. Determining a recovery state;
A program that causes a computer to execute.   A recording medium on which the program according to claim 7 is recorded.

Images