CN101114955A - Jitter detection based congestion control method in city domain Ethernet - Google Patents

Abstract

The invention relates to a transmission control method of the multimedia data, in particular to a congestion control method based on the jitter measurement in the metro Ethernet network. The method utilizes the priority bit position of the C-VLAN Tag of the Q-in-Q the metro Ethernet network as the counter of the node time delay jitter. The accumulated time delay jitter of the multimedia data frame in the transmission process from end to end is followed. The data of the small jitter corresponds to the higher priority and the data of the higher jitter corresponds to the lower priority. The Ethernet data frame, the jitter value of which is larger than the user tolerate threshold value, is lost by the determination of the user tolerate threshold value. The invention can effectively reduce the average time delay jitter of the multimedia flow, which can improve the good throughput of the multimedia service and the service quality of the multimedia service.

Description

Congestion control method based on jitter detection in metro Ethernet

Technical Field

The invention relates to a multimedia data transmission control method, in particular to a congestion control method based on jitter detection in a metropolitan area Ethernet.

Background

With the wide application of IPTV, video monitoring, triple play, etc., the multimedia video service is one of the fastest growing applications in the current metropolitan area network. In the current metropolitan area network domain, ethernet has become one of the very broad solutions. In fact, many operators currently implement ethernet services on metropolitan networks. How to better carry multimedia applications in metro ethernet is one of the issues that needs to be solved by current metro ethernet networks.

When a video stream or an audio stream for a multimedia application is transmitted on a packet-switched network node, it is generally assumed that a constant delay relationship is maintained between data packets, but in practice this is difficult. In the actual transmission process, the occurrence of congestion or the difference of transmission delay causes the data packets to have a variation in network delay when reaching the destination, that is, delay jitter is generated, which causes the service quality of the multimedia data stream to be degraded. If the resulting delay jitter is too large when multimedia data streams traverse the network, the user will become unavailable to the user even if they receive the data streams. This is because it is difficult to re-align the timing of data packets between a data stream or multiple data streams at the user end. In some multimedia systems [1] [2], a buffer is allocated at the client to solve the problem. However, such buffers merely increase the tolerance of the network to delay jitter and do not substantially solve the problem.

This situation becomes even worse if the accumulated delay jitter of a large number of multimedia data packets during network transmission exceeds the tolerance of the client. In this case, although these multimedia data are not available to the user, they still occupy the bandwidth of the network, and the transmission of these useless data certainly increases the congestion level of the network. This new congestion problem cannot be solved by conventional queue management methods such as Random Early Detection (RED) 3 or drop tail algorithm (DropTail) 3.

The main representatives of conventional congestion control methods are Random Early Detection (RED) and missing tail arithmetic (DropTail) [3]. RED is an active queue management mechanism proposed by the IETF. The working principle of the method is that according to the current queue length and the threshold value of the queue size, incoming packets are randomly discarded with a certain probability. The effectiveness of RED has been proven by some practices, but still has some drawbacks. For example, the performance of RED algorithms is sensitive to design parameters and network conditions, which can cause queue oscillations, throughput degradation and delay jitter to be exacerbated under certain network loading conditions [4]. RED generally does not provide any protection mechanism for TCP flows. When there are UDP flows and their shared queues, UDP traffic will typically be larger than TCP traffic. How to achieve TCP friendliness in RED algorithms is a hot topic in the field of RED research. In document [5], a variant of the RED algorithm is presented which considers the packet size at the time of discarding to achieve better results and fairness. In document [6], the loss probability in RED is calculated using the average delay instead of the average queue length. In document [7], a preferential drop strategy is proposed to identify individual data streams that occupy excessive bandwidth, by which the rates of these streams are compressed within a specified range of values. Other variant algorithms for large amounts of RED have been generated in order to improve and perfect the deficiencies and defects of the RED algorithm, and the most significant are WRED [8], stabilized-RED [9], self-configuration RED [10], adaptive RED [11], FRED [12] and Balanced-RED [13], wherein FRED and Balanced-RED are mainly used to solve the fairness problem of RED, and the rest are used to enhance the stability of RED. None of these conventional congestion control methods adequately takes into account the characteristics of network traffic, especially ignoring some of the characteristics of multimedia traffic that are different from other data traffic, such as the sensitivity of multimedia traffic to delay jitter. Data that, although transmitted to the client, has accumulated more than the user-tolerated delay jitter is of virtually no significance to the user. Conventional congestion control methods are not capable of this situation.

In the current metropolitan area network field, ethernet has been one of the hottest solutions. In fact, many operators currently implement ethernet services on metropolitan networks. However, the extension of ethernet technology to metropolitan area networks entails a number of troublesome problems, such as breaking through the limit of 4096 VLANs (virtual local area networks), transparent LAN traffic connections, quality of service guarantees, and so on. In order to break the limit of 4096 VLANs and to enable transparent LAN traffic transmission in metro ethernet networks, IEEE802.1ad (Provider Bridge) standard for Q-in-Q was defined by the IEEE 2005 [14]. Q-in-Q is also called Stack VLAN, and the expansibility of the network is improved by adding a layer of Qtag. In Q-in-Q technology, the outer VLAN tag is called S-VLAN, which belongs to the operator network wide VLAN space, while the inner VLAN tag is called C-VLAN, which belongs to the client VLAN space. When the operator receives the Ethernet frame sent by the user at UNI, an S-VLAN Tag is added to the Ethernet frame marked by the IEEE 802.1Q Tag of the user. The Q-in-Q Ethernet frame format of IEEE802.1ad is shown in FIG. 1.

The S-VLAN Tag is embedded after the ethernet destination MAC address and source MAC address, and before the C-VLAN. In an 802.1Q VLAN, the priority in the VLAN takes three bits and supports eight different priorities to differentiate between different services. When the metropolitan area is used for forwarding data frames in the Ethernet, only priority setting in S-VLAN is generally considered, and setting of S-VLAN ID and S-VLAN CoS is commonly used for identifying service and service performance (CoS). Whereas the C-VLAN Tag is typically not changed nor accessed.

Disclosure of Invention

The invention aims to provide a congestion control method based on jitter detection in a metro Ethernet network aiming at the defects of the prior art, and the method can effectively reduce the average delay jitter of multimedia stream and improve the effective throughput of multimedia service, thereby improving the service quality of the multimedia service.

The technical scheme of the invention is as follows: a congestion control method based on jitter detection in a metro Ethernet utilizes a priority bit in a C-VLAN Tag of Q-in-Q in the metro Ethernet as a counter of node delay jitter, the delay jitter accumulated in the end-to-end transmission process of a multimedia data frame is tracked, data with a small jitter value corresponds to a high priority, data with a large jitter value corresponds to a low priority, and the Ethernet data frame with the jitter value larger than a user tolerance threshold is discarded through the determination of the user tolerance threshold.

The congestion control method based on jitter detection in metro ethernet network as described above, wherein the jitter local _ jitter experienced by the ethernet data frame in the switch is obtained by the following formula:local _ jitter = delay-ave _ delay, where delay is a delay experienced by the ethernet data frame in the switch, and ave _ delay is an average delay experienced by the data frame in the switch; the time delay is calculated by the following equation: delay = buf _ oc _ Packet _ size/link _ bw, in the formula, buf _ oc is the current buffer occupancy of the output queue, link _ bw is the available link bandwidth, and Packet _ size is the Packet length; the average delay ave _ delay is calculated by an exponentially weighted average of the delays per data frame, with the formula ave _ delay = ave _ delay (1-W) _d )+delay*W _d ，W _d Are weights.

In the congestion control method based on jitter detection in the metro ethernet network, when a drop decision of a packet is made, the jitter value total _ jitter experienced by an ethernet data frame from a source to a current node is a local jitter value plus the value of a jitter counter corresponding to a priority in a C-VLAN Tag carried in the ethernet data frame, and when total _ jitter > user _ threshold/2 or total _ jitter < -user _ threshold/2, the ethernet frame is dropped, (-user _ threshold/2, user _ threshold/2) is a tolerance threshold range of a user; if the total _ jitter is within the user tolerance threshold, then the jitter counter is updated and the value of counter is converted to the corresponding priority in the C-VLAN Tag.

The congestion control method based on jitter detection in the metro ethernet network as described above, wherein the jitter value is divided into 8 different levels, positive value 4 levels and negative value 4 levels according to the threshold that the user can tolerate.

In the congestion control method based on jitter detection in the metro ethernet network, the lowest bit of the three bits of the priority in the C-VLAN Tag is used to represent the positive and negative jitter values, 0 represents the positive jitter value, and 1 represents the negative jitter value.

In the congestion control method based on jitter detection in the metro ethernet network, the two high bits in the priority respectively represent 4 levels of jitter positive values or negative values, that is, from level 0 to level 3, the binary bit corresponding to level 0 is 11, the binary bit corresponding to level 1 is 10, the binary bit corresponding to level 2 is 01, and the binary bit corresponding to level 3 is 00.

The method provided by the invention detects the time delay jitter of the multimedia data stream at the network node, and utilizes the Q-in-Q C-VLAN Tag in the metro Ethernet as a counter of the node time delay jitter, and improves the service quality of the multimedia service by discarding the data frames exceeding the tolerance threshold of the user. The result shows that compared with the traditional congestion control methods of RED and DropTail, the method not only can keep the friendliness of TCP, but also can more effectively reduce the average delay jitter of the multimedia stream and improve the effective throughput of the multimedia service.

Drawings

FIG. 1 is an Ethernet frame format with Q-in-Q VLAN tag.

Fig. 2 is a corresponding relationship between the C-VLAN Tag priority and the jitter value.

FIG. 3 is a simulation test topology of the present invention.

Fig. 4 is a diagram illustrating the throughput of TCP flows with different frame lengths.

Fig. 5 is a diagram illustrating the throughput of UDP streams with different frame lengths.

Fig. 6 is a scale graph of UDP and TCP throughput for different frame lengths.

Fig. 7 is a graph comparing the average delay jitter of a frame length of 128 bytes.

Fig. 8 is a graph comparing the average delay jitter of a frame length of 512 bytes.

Fig. 9 is a graph comparing the average delay jitter of a frame length of 1024 bytes.

Fig. 10 is a graph comparing the effective throughput for a frame length of 1024 bytes.

Fig. 11 is a structural diagram of a line card switching chip in an S4606 switch implementing the present invention.

Fig. 12 is a block diagram of a software system architecture for applying the method of the present invention.

Detailed Description

The invention is described in detail below with reference to the figures and examples.

The basic idea of a congestion control method based on Jitter Detection (JDCC) is to detect and track the delay jitter during the transmission of multimedia data streams, and if the delay jitter exceeds the tolerance threshold of the user, the data packet is discarded, which reduces unnecessary bandwidth waste. In order to keep track of the delay jitter accumulated during the end-to-end transmission of multimedia data frames, a jitter counter is required. It is obvious that the counter is to record the jitter experienced during the whole transmission in a certain field in the data frame. In the current metro ethernet network, a Q-in-Q networking mode is very common, and in the Q-in-Q networking mode, after entering an operator network, forwarding decision and priority policy are both for an S-VLAN Tag, while three bits of priority of a C-VLAN Tag are unused, and only after reaching a user end, the C-VLAN Tag is activated, and there is little possibility that the user end has basically been congested, and at this time, the function of a jitter counter may also be stopped. Therefore, the priority of the C-VLAN Tag is used as a jitter value counter in the Ethernet frame transmission process in the method of the invention.

The priority in the C-VLAN Tag is not used in the transmission process of the operator network, so the JDCC method adopts three priority bits in the C-VLAN Tag as a time delay jitter counter. However, when the data frame enters the network node of the user end, the S-VLAN Tag is stripped, the C-VLAN Tag plays a role, and the priority in the C-VLAN Tag is also a counter of a jitter value at the same time. For multimedia data streams, those data with smaller jitter values should be forwarded first and therefore correspond to higher priority, while for those data frames with already large jitter values, it may not be meaningful for the user and should be discarded first when congestion occurs and therefore correspond to lower priority. It follows that the correspondence between jitter values and priorities is such that a higher jitter value corresponds to a lower priority and a lower jitter value corresponds to a higher priority. In the invention, the lowest bit of three bits of the jitter value is used for representing the positive and negative of the jitter value, 0 represents the positive jitter value, and 1 represents the negative jitter value. The jitter value is divided into 8 different grades according to the tolerable threshold of the user, wherein the positive value is 4 grades, and the negative value is 4 grades. The upper two bits in the priority represent 4 levels of jitter positive or negative values, i.e., from level 0 to level 3, respectively. Since the higher the jitter value, the lower the priority, and the lower the jitter value, the higher the priority. So the binary bit corresponding to level 0 is 11; a level 1 corresponds to a binary bit of 10, a level 2 corresponds to a binary bit of 01, and a level 3 corresponds to a binary bit of 00. Such as: if the jitter value is in positive rank 0, then the corresponding priority is: the upper two bits are bit 1 and the lowest bit is 0, i.e. the corresponding priority is 6. If the jitter value is negative in level 0, the queue should have a priority of 7. Whereas if the jitter value is high, falling in level 3, the corresponding priority is 0 or 1. From the above analysis, it can be seen that the value of the priority is just the inverse of the jitter value level. The specific correspondence can be seen in fig. 2.

If the C-VLAN Tag of the user data itself carries a priority before the user data enters the operator network, the priority in the C-VLAN Tag of the user data may be mapped to the S-VLAN Tag. Thus the priority in the C-VLAN Tag can be used to identify the level of the jitter field value. If the C-VLAN Tag of the user data before entering the operator network does not have priority, the priority is distributed by adopting the method. The specific implementation code is as follows:

For each classified packet arrived: (1) Estimate the output link capacity:link_bw (2) Update the current buffer occupancy:buf_oc； (3) Calculate the delay:delay＝buf_oc*Packet_size/link_bw； (4) Calculate the average delay: ave_delay＝(1-W d )*ave_delay+W d *delay； local_jitter＝delay-ave_delay (5) Calculate the total jitter: step＝user_th/8； if(prio is even)//(0，2，4，6 positive value){ counter = NOT (prio > 1); (two high order negation) total_jitter＝local_jitter+counter*step； } else//(prio＝1，3，5，7){ counter = NOT (prio > 1) (two high bits negated) total_jitter＝local_jitter-counter*bound； } (6) if(-user_th/2＜total_jitter＜user_th/2) { queue the packet； update the counter； if(total_jitter＞0) { temp＝8*total_jitter/user_th； counter＝floor(temp)； prio＝NOT(counter)； prio＝prio＜＜1； } else { temp＝8*total_jitter/user_th； counter＝ceil(temp)； prio＝NOT(|counter|)； prio＝(prio＜＜1)|1； } } else { discard the packet； } (7) write prio to C-VLAN Tag

The above procedure describes pseudo-code of the JDCC method, the variable delay being the delay experienced by an ethernet data frame in the switch, which can be estimated by: current buffer occupancy of the output queue (buf _ oc) and available link bandwidth (link _ bw):

delay＝buf_oc*Packet_size/link_bw； (1)

the average delay experienced by a data frame in a switch is calculated by an exponentially weighted average of the delay of each data frame, as shown in equation (2):

ave_delay＝ave_delay*(1-W _d )+delay*W _d (2)

by exponentially weighted averaging, the increase in short term average delay can be effectively smoothed out. The jitter experienced by the ethernet data frame in the switch is then given by the following equation:

local_jitter＝delay-ave_delay； (3)

the jitter value total _ jitter experienced by the Ethernet data frame from the source to the current node is the local jitter value plus the value of the jitter counter corresponding to the priority in the C-VLAN Tag carried in the Ethernet data frame. If total _ jitter exceeds the tolerance threshold of the user, then:

total_jitter＞user_threshold/2 or total_jitter＜-user_threshold/2

(4)

the ethernet frame is discarded. If the total _ jitter is within the tolerance threshold of the user, the jitter counter is updated, and the counter is converted into the corresponding priority in the C-VLAN Tag according to the corresponding relation in FIG. 2.

The performance of the JDCC method can be obtained by measuring the delay jitter of the received data packets, using the sample average delay

Due to the use of a limited number of sample packets N _s . By using the Cheyshev inequality, the upper probability limit P (drop) for dropping a packet can be derived. Suppose packet D is received _i Are independent of each other with a standard deviation sigma _d And (5) the distribution is the same. Sample average time delayComprises the following steps:

the Chebyshev inequality can be used to get a loose upper bound, the probability of dropping a packet P (drop), and the delay of receiving a packet D _i And average time delay

By the quantity N _s And a discard threshold. The discard threshold user _ th is defined as:

user_th＝2a _j J _max ； (6)

J _max is the maximum delay jitter allowed, a _j E (0, 1). The drop probability P (drop) is then:

available according to the chebyshev inequality:

equation (7) can be written as:

from equation (8), the maximum tolerance threshold or the maximum delay jitter value of a given user can obtain an upper limit of the dropping probability, and from equation (8), it can be seen that the higher the dropping threshold value is, the smaller the dropping probability is. Meanwhile, the upper performance limit of the drop probability is also given by the formula (8).

Figure 3 gives the topological graph of the experiment. In the topology diagram, the host is connected to the operator network through a user access switch (customer switch in the diagram). The data flow sent from the user host is not provided with VLAN Tag, after the data flow enters the user switch, the C-VLAN Tag is marked on the port of the user switch connected with the user host, and the C-VLAN is distributed according to the port. The VLAN IDs of the subscriber hosts A1-A4 in the test are 100-400 respectively, but have no priority. The VLAN TRUNK mode is configured on the uplink port of the user switch. Thus, the data stream from the user uplink port is labeled with the C-VLAN Tag. After entering the SP switch, the SP switch assigns S-VLAN Tag to the data stream according to the port, and the user data stream becomes Q-in-Q mode after entering the SP network. In the experiment, the SP VLANs of the ports on the SP switches connected to the two private branch exchanges were 1000 and 2000, respectively. Since the JDCC method of the present invention is run in an SP network, the simulation test is mainly concerned with the operator network part, i.e. the SP Networks part in fig. 3.

The data streams of TCP (streaming using FTP) and UDP (multimedia streaming, such as MPEG-2 video or audio, with a bit rate of 6 Mb/s) are transmitted on the two subscriber hosts connected to each subscriber exchange. The link capacity from the private branch exchange to the SP exchange is set to 10Mb/s and the transmission delay is set to 3ms. The samples of the MPEG-2 video stream are in VBR format. For simplicity of the experiment, UDP streams are used in the following discussion to represent video streams.

In the experiment the data frame would traverse a link with 3 SP switching nodes and then reach the customer premises switch at the other end. The nodes between the SP switches are interconnected by 10Mb/s bandwidth links with a propagation delay of 2ms. The 3 SPsThe link between the switches is a bottleneck over the entire link. RED and JDCC methods are implemented on SP switching nodes for congestion control of TCP and UDP flows. Setting the limit of the data frames in the queue on each node to be 150 in the experiment; the test has the advantages that the test has 256, 512 and 1024 frames, and the test time is 10 seconds. The rate of each stream for UDP and TCP is set to 6Mbps. The threshold of JDCC method is set to 0.1s, weight W _d Set to 0.02; the parameters in the RED method are set to a maximum threshold of 100 frames and a minimum of 30 frames. In tests, performances of the JDCC method and the traditional congestion control methods RED and DropTail are compared and analyzed mainly from three aspects of TCP friendliness, delay jitter and effective throughput of multimedia streams.

UDP is a connectionless protocol, which has no corresponding traffic control mechanism and does not deal with packet loss. When the UDP stream and the TCP stream share the bandwidth, this characteristic of the UDP stream greedily occupies a large amount of bandwidth, and reduces the throughput of the TCP stream. Therefore, it is important that the JDCC method proposed by the present invention has certain TCP friendliness, and it should not reduce the throughput of TCP stream and achieve the same friendliness as the traditional RED, dropTail method. In the experiment, before placing TCP and UDP streams in FIFO queues, different queue management policies are applied to them, respectively. The test results in fig. 4 show the average throughput of TCP flows for different data frame sizes and different queue management strategies. From the test results, it can be seen that the JDCC method can achieve TCP throughput similar to RED and DropTail compared to the conventional congestion control method. While the throughput of the UDP streams for different data frame sizes is given in fig. 5, the same experiment as in fig. 4 is used. The average throughput of UDP using the JDCC method decreases as the number of data frames increases, while it is smaller than the DropTail method and the RED method. This is because some UDP packets accumulate a large amount of delay jitter and are discarded. The TCP friendliness of the method is directly related to the ratio of UDP throughput to TCP throughput. Fig. 6 gives the ratio between UDP and TCP throughput at different frame lengths. It can be seen from experiments that the JDCC method is close in TCP friendliness compared to other mature congestion control methods.

The delay jitter is a very important parameter index for multimedia services, and ensuring smaller delay jitter is an important means for improving the service quality of the multimedia services. The delay jitter of a multimedia data stream using the RED method is defined as follows: j. the design is a square _i = delay-ave _ delay. Drawing (A)Fig. 7, fig. 8, and fig. 9 respectively show a comparison graph of average values of delay jitters of the three methods in the case of different frame lengths. It can be seen from the figure that the JDCC method has a significantly lower average delay jitter value than the conventional RED and DropTail methods for various frame lengths. It can also be seen that the three methods, including DropTail, RED and JDCC, all have the same trend, and this periodic trend is mainly caused by the traffic control mechanism of the TCP protocol itself. Test knot in the figureThe JDCC method provided by the invention can achieve lower delay jitter compared with DropTail and RED methods.

UDP streams of multimedia are very sensitive to delay jitter. In the case where the delay jitter of a packet is greater than the tolerance threshold of the user, it is actually meaningless for the user even if the packet is successfully delivered to the client. And if there is not enough bandwidth, which would increase the likelihood of congestion, forwarding these packets with larger delay jitter at the switching node would reduce the available bandwidth of the transmission channel. But also causes some other data frames to add jitter to the delay, causing more data frames to become meaningless data for the user. Data frames exceeding the user tolerance threshold are dropped in the JDCC method.

In fact, during the transmission of data, the output useful to the user is of interest, not just the throughput. By "valid" it is meant that the multimedia data frame arrives at the client with jitter tolerance, i.e. the data frame arriving at the client has jitter < user _ threshold. A "valid" frame of multimedia data suffices to reach its position before being played to the user. The user tolerance threshold is set in the range of 0.03-0.1 seconds during the trial. The results of the experiments are shown in fig. 10 and show that JDCC method can achieve higher average goodput than DropTail method and RED method. The entire trend is clearly seen in fig. 10, which shows that the experimental results are consistent with the theoretical analysis.

The hardware platform adopted by the JDCC method is mainly based on an S4606 switch. The JDCC method is mainly implemented on LC-48FE promiscuous line cards (48 x100-BASET line cards).

The JDCC method of the present invention is mainly implemented on the switching chip of the LC-48FE hundred mega line card, the structure of the switching chip is shown in fig. 11, and the line card switching core in S4606 mainly includes the following parts:

1) Packet Parser (Packet Parser): the main function is to analyze the incoming data packet;

2) Packet Modifier (Packet Modifier): mainly used for modifying the next hop MAC and other functions of the data packet forwarded by the L3;

3) The CPU management interface controller (CMIC: CPU Management Interface Controller): mainly responsible for the interface with CPU part; including PCI interfaces (PCI interfaces) and Management counters (Management counters);

4) Packet Filter Processor (Packet Filter Processor): the filtering function of the data packet is mainly completed;

5) Memory Management Unit (Memory Management Unit): the method is mainly responsible for cache management and flow management in the exchange chip; the method is divided into two parts, namely a Buffer Management unit (Buffer Management) and a Traffic Management unit (Traffic Management). Buffer Management unit (Buffer Management): the JDCC algorithm which mainly completes the functions of management, access control and the like of a cache queue in the switching chip is completed in the part; traffic Management unit (Traffic Management): the method mainly completes the functions of limiting the speed of the flow, shaping, managing and scheduling queues and the like.

6) Layer two forwarding logic (L2 Switching): the method mainly completes the two-layer forwarding of the data packet;

7) Three-tier forwarding logic (L3 Switching): mainly completes the forwarding of three layers (IP layers) of data packets.

The software platform of the invention is an embedded operating system and a TCP/IP protocol stack on the upper layer; the embedded operating system adopts a VxWorks5.5 kernel; the protocol stack part mainly adopts a USP unified software platform developed by a beacon network company; the software development platform adopted the version of tornado2.2 for PPC. The system structure of the whole software is shown in fig. 12. The description in fig. 12 refers to english abbreviations:

ARP (Address Resolution Protocol): an address resolution protocol;

BGP (Border Gateway Protocol): a border network management routing protocol;

CLI (Command Line Interface): a command line;

console: a console serial port;

DCP (Distributed Communication Protocol): a distributed communication protocol;

DHCP (Dynamic Host Configuration Protocol): a dynamic host configuration protocol;

FTP (File Transmission Protocol): a file transfer protocol;

GARP (General Attribute Register Protocol): generic attribute registration protocol

GMRP (GARP Multicast Register Protocol): GARP multicast register protocol;

GVRP (GARP Vlan Register Protocol): GARP Vlan registration protocol;

HA (High Availability): high availability;

IDB (Information Database): an information database;

ICMP (Internet Control Message Protocol): an Internet control message protocol;

IGMP (Internet Group Management Protocol): an internetwork group management protocol;

IP (Internet Protocol): an internet protocol;

MSTP (Multiple spinning Tree Protocol): multiple spanning tree protocol

OSPF (Open Shortest Path First): a shortest path first protocol;

PIM/SM (Protocol Independent Multicast/spark Mode): a sparse mode independent multicast protocol;

RIP (Routing Information Protocol): a routing information protocol;

RSTP (Rapid breathing Tree Protocol): a rapid spanning tree protocol;

SNMP (Simple Network Management Protocol): a simple network management protocol;

STP (spinal Tree Protocol): a spanning tree protocol;

TCP (Transmission Control Protocol): a transmission control protocol;

telnet: a remote terminal protocol;

TFTP (Trivisual File Transmission Protocol): a simple file transfer protocol;

UDP (User Datagraph Protocol): user datagram protocol.

Reference documents

[1]S.P.Chan and C.W.Kok，“Protocol and buffer design for multimedia-on-demand system，”in Proc.IEEE PCM，Oct.2001，pp.1010-1015.

[2]，“Bitrate adaptation flow control for multimedia-on-demand，”in Proc.IEEE ICC，May 2002， pp.2503-2507.

[3]S.Floyd and V.Jacobson，“Random early detection gateways for congestion avoidance，” IEEE/ACM Trans.Netw.，vol.1，no.4，pp.397-413，Aug.1993.

[4]S.Floyd，A report on some recent developments in TCP congestion control.IEEE Communication Magazine，April 2001.

[5]S.De Cnodder，O.Elloumi，and K.Pauwels，“RED behavior with different packet sizes，”in Proc.5th IEEE Symp.Computers Commun.，2000，pp.793-799.

[6]J.B.Pippas and I.S.Venieris，“A RED variation for delay control，”in Proc.ICC，vol.1，2000， pp.475-479.

[7]S.Floyd and R.Mahajan，“Controlling High-Bandwidth Flows at the Congested Router，”，Nov. 20，2000.Tech.Rep.of ACIRI.

[8]Cisco system.Distributed weighted random early detection.http://cco.cisco.com

[9]Ott Teunis J，Lakshman T V，Wong Larry H.SRED：Stabilized RED.In IEEE INFORCOM’99 New York，USA，1999.1346～1355.

[10]Feng W，Kandlur D，Saha D，Shin K.A self-configuring RED gateway.In Proc.IEEE INFOCOM’99.New York，USA，1999.1320～1328

[11]Floyd S，Gummadi R，Shenker S.Adaptive RED：An algorithm for increasing the robustness of RED’s active queue management.http://www.cs.berkeley.edu/

[12]Lin D，Morris R.Dynamics of random early detection.In Proceedings of the ACM SIGCOMM Conference on Applications，Technologies，Architectures，and Protocols for Computer Communications.New York，USA，1997.127～138

[13]Anjum F，Tassiulas L.Balanced-RED：Analgorithm to achieve fairness in Internet.In Proceedings of IEEE INFOCOM’99.New York，USA，1999.

[14]IEEE Std 802.1ad ^TM -2005，IEEE Standard for Local and Metropolitan Area Networks：Virtual Bridged Local Area Networks-Amendment 4：Provider Bridges.

Claims (8)

1. A congestion control method based on jitter detection in a metro Ethernet is characterized in that: the method utilizes a priority bit in a C-VLAN Tag of Q-in-Q in a metropolitan area Ethernet as a counter of node delay jitter, the delay jitter accumulated in the end-to-end transmission process of a multimedia data frame is tracked, data with small jitter values correspond to high priority, data with large jitter values correspond to low priority, and the Ethernet data frame with the jitter values larger than a user tolerance threshold is discarded through determining the user tolerance threshold.

2. The congestion control method based on jitter detection in a metro ethernet network according to claim 1, characterized in that: the jitter local _ jitter experienced by the ethernet data frame in the switch is obtained by the following equation: local _ jitter = delay-ave _ delay, where delay is the delay experienced by the ethernet data frame in the switch, and ave _ delay is the average delay experienced by the data frame in the switch.

3. A method for congestion control in a metro ethernet network based on jitter detection according to claim 2, characterized in that: the time delay is calculated by the following equation: delay = buf _ oc _ Packet _ size/link _ bw, in the formula, buf _ oc is the current buffer occupancy of the output queue, link _ bw is the available link bandwidth, and Packet _ size is the Packet length; the average latency ave _ delay is calculated by an exponentially weighted average of the latencies of each data frame, with the formula ave _ delay = ave _ delay (1-W) _d )+delay*W _d ，W _d Are weights.

4. The congestion control method based on jitter detection in a metro ethernet network according to claim 1, characterized in that: when the discarding judgment of the data packet is carried out, the jitter value total _ jitter experienced by the Ethernet data frame from the source to the current node is the local jitter value plus the value of the jitter counter corresponding to the priority in the C-VLAN Tag carried in the Ethernet data frame.

5. The congestion control method in a metro ethernet network based on jitter detection according to claim 4, wherein: if total _ jitter > user _ threshold/2 or total _ jitter < -user _ threshold/2, then the Ethernet frame is discarded, (-user _ threshold/2, user _ threshold/2) being the tolerance threshold range for the user; if total _ jitter is within the user tolerance threshold, the jitter counter is updated and the value of counter is converted to the corresponding priority in the C-VLAN Tag.

6. A method of congestion control in a metro ethernet network based on jitter detection according to claim 1 or 2 or 3 or 4 or 5 characterized in that: the jitter value is divided into 8 different levels according to the threshold which can be tolerated by the user, wherein the positive value is 4 levels, and the negative value is 4 levels.

7. The congestion control method based on jitter detection in a metro ethernet network according to claim 6, characterized in that: the lowest bit of three bits of the priority in the C-VLAN Tag is used for representing the positive and negative of the jitter value, 0 represents the positive value of the jitter, and 1 represents the negative value of the jitter.

8. A method for congestion control in a metro ethernet network based on jitter detection according to claim 7 characterized in that: the high two bits in the priority respectively represent 4 levels of the jitter positive value or negative value, that is, from level 0 to level 3, the binary bit corresponding to level 0 is 11, the binary bit corresponding to level 1 is 10, the binary bit corresponding to level 2 is 01, and the binary bit corresponding to level 3 is 00.

Images