CN102761470B - Multipath TCP (transmission control protocol) message scheduling method - Google Patents

Abstract

The present invention provides a multipath TCP transmission protocol message scheduling method, combining subflow packet loss rate and difference in time delay, reasonably scheduling subflow messages, and solving the bottleneck subflow problem in multipath; Each sub-stream is transmitted together, making full use of the network bandwidth, effectively improving the network throughput and the stability and robustness of the transmission.

Description

A Multipath TCP Transmission Protocol Packet Scheduling Method

technical field

The invention relates to the technical field of multipath reliable transmission, in particular to a multipath TCP transmission protocol message scheduling method.

Background technique

Traditional TCP is a connection-oriented reliable transport protocol. Its congestion control ensures the friendliness between TCP streams. However, this congestion mechanism believes that packet loss is caused by network congestion. In the case of high packet loss rate and delay In a wireless network, the transmission performance of traditional TCP will be severely degraded due to large packet loss and delay, which cannot meet the transmission requirements of end-to-end users.

The IETF proposed the multipath TCP transmission protocol (Multipath TCP) in 2010. Multipath TCP is an extension of traditional TCP. It improves the throughput of the communication terminal by increasing the number of paths (that is, the number of subflows), divides the received data flow into multiple subflows, and then separates different subflows through multiple paths. transmitted to the peer node. The multi-path protocol mechanism is transparent to upper-layer applications. Through multiple paths for joint transmission, network resources are fully utilized, network throughput is improved, and transmission stability and robustness are effectively guaranteed.

In the existing multi-path TCP protocol, the round-robin scheduling method is used to schedule and allocate packets when the link is idle, without considering the differences between sub-flows, due to the sub-flow link bandwidth and packet loss rate The actual throughput of each sub-flow is different during the transmission process, and the poor sub-flow will affect the overall throughput during the transmission process and become a bottleneck sub-flow.

Therefore, a technical problem that needs to be solved urgently by those skilled in the art is: how to innovatively propose a multi-path TCP transmission protocol message scheduling method to solve the problems existing in the prior art, according to the link of each sub-flow Real-time adjustment of sub-flow messages, so that each sub-flow can jointly complete the transmission task according to the actual transmission capacity, solve the problem of bottleneck sub-flow, and improve the throughput, stability and robustness of transmission.

Contents of the invention

The technical problem to be solved by the present invention is to provide a multi-path TCP transmission protocol message scheduling method to solve the transmission problem caused by the difference between multi-paths and the impact of the bottleneck link caused by the poor sub-flow on the overall throughput .

In order to solve the above problems, the present invention discloses a method for scheduling multipath TCP transmission protocol messages, the method comprising:

After the sub-flow connection is established, the link information of each sub-flow is collected in real time; the link information includes packet loss rate, actual transmission delay and current congestion window;

Calculate the theoretical transmission delay of each sub-flow, and sort the packet loss rate and theoretical transmission delay of all sub-flows;

Packets are scheduled according to the theoretical transmission delay, packet loss rate, and congestion window of each subflow.

Preferably, the scheduling of the message according to the theoretical transmission delay, packet loss rate and congestion window of each sub-flow specifically includes:

A. Sort the sub-flows according to the theoretical transmission delay and packet loss rate of the sub-flows, and determine the packet scheduling priority of the sub-flows;

B. Determine the number of packets required for each subflow according to the congestion window and the number of confirmed packets;

C. Distribute the number of packets required by the current sub-flow to each sub-flow in turn according to the priority of the sub-flow, and at the same time, determine the link quality of each sub-flow in real time:

If it is normal, continue to allocate the required number of packets;

Otherwise, repeat steps A, B and C.

Preferably, the link quality is judged by substream theoretical transmission delay and packet loss rate.

Preferably, if the packet loss rate is 1, the actual transmission delay p and the current congestion window w, then during the transmission process, the theoretical arrival time of each message is:

$\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{<span\; class="notranslate">\; \&infin;</span>}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; kp</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\frac{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; p</span>}}{{\mathrm{<span\; class="notranslate">\; l</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; .</span>$

Compared with the prior art, the present invention has the following advantages:

The scheme of the present invention combines the differences in the packet loss rate and time delay of the sub-flows, and reasonably schedules the packets of the sub-flows, which solves the problem of the bottleneck sub-flow in multipath; all sub-flows are transmitted together, making full use of the network bandwidth, Effectively improve the stability and robustness of network throughput and transmission.

Description of drawings

Fig. 1 is a flow chart of a method for scheduling multipath TCP transmission protocol packets according to an embodiment of the present invention;

Fig. 2 is a schematic diagram of a multipath TCP transmission protocol message scheduling method according to an embodiment of the present invention.

Detailed ways

In order to make the above objects, features and advantages of the present invention more comprehensible, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

In a wireless network, due to interference, packet loss, and high delay, TCP frequently disconnects and rebuilds, which affects the stability and robustness of transmission, and the network throughput is low, which cannot meet user needs. The existing multipath TCP introduces multipath to improve throughput, but due to the differences in subflows, poorer links may become transmission bottlenecks, affecting the transmission of better subflows, failing to achieve ideal throughput, or even appearing Deadlock: When the receiving window is full, the better sub-flow waits for the inferior sub-flow to complete the receiving window before sliding forward.

The scheme of the present invention combines the differences in the packet loss rate and time delay of the sub-flows, and reasonably schedules the packets of the sub-flows, which solves the problem of the bottleneck sub-flow in multipath; all sub-flows are transmitted together, making full use of the network bandwidth, Effectively improve the stability and robustness of network throughput and transmission.

Example:

With reference to Fig. 1, show the flow chart of a kind of multipath TCP transmission protocol message scheduling method of the present invention, described method specifically comprises:

Step S101, after the sub-flow connection is established, link information of each sub-flow is collected in real time; the link information includes packet loss rate, actual transmission delay and current congestion window;

Step S102, calculating the theoretical transmission delay of each sub-stream, sorting the packet loss rate and theoretical transmission delay of all sub-streams;

Step S103, scheduling packets according to the theoretical transmission delay, packet loss rate and congestion window of each subflow.

Concrete step S103 comprises the following steps:

A. Sort the sub-flows according to the theoretical transmission delay and packet loss rate of the sub-flows, and determine the packet scheduling priority of the sub-flows;

B. Determine the number of packets required for each subflow according to the congestion window and the number of confirmed packets;

C. Distribute the number of packets required by the current sub-flow to each sub-flow in turn according to the priority of the sub-flow, and at the same time, determine the link quality of each sub-flow in real time:

If it is normal, continue to allocate the required number of packets;

Otherwise, repeat steps A, B and C.

The method described in the present invention is introduced through specific implementation below. The present invention provides a multipath TCP protocol message scheduling method, which mainly solves the transmission problem caused by the difference between multipaths and the bottleneck link caused by poor subflows. impact on overall throughput. In multi-path TCP transmission, sub-flow link quality: bandwidth, packet loss rate, delay and congestion window are different.

Specifically, the following concepts are used to measure sub-flows and schedule them according to their actual conditions:

During transmission, each subflow updates its own link information in real time: including packet loss rate 1, actual transmission delay p, and current congestion window w. Since the loss rate is different, the actual transmission time of the message should take the packet loss rate into consideration. During the transmission process, the theoretical arrival time of each message is:

$\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{<span\; class="notranslate">\; \&infin;</span>}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; kp</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\frac{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; p</span>}}{{\mathrm{<span\; class="notranslate">\; l</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}$

The theoretical transmission delay d comprehensively considers the real-time packet loss rate of the link and the transmission delay, which reasonably reflects the time required to successfully transmit a message.

Calculate the theoretical transmission delay according to the real-time link information of each subflow in multipath TCP. If multipath TCP wants to achieve the maximum throughput, it should make full use of the subflow bandwidth, which requires avoiding the bottleneck subflow problem as much as possible, and reducing the impact of poor links on better subflows.

During the multipath scheduling process, the main factors to be considered are: the theoretical transmission delay of the sub-flow and the current window size. The transmission delay determines the sub-flow priority, and the window size determines the number of packets allocated to the sub-flow.

Define the scheduling factor as the priority of the sub-flow and the required message data, represented by [d _i , Δw _i ], where d _i determines the priority of the sub-flow, and the sub-flow with higher priority can get the first message, Δw _i represents the number of packets currently required by the subflow.

The message scheduling method of multi-path TCP, its realization schematic diagram can refer to Fig. 2, specifically comprises the following steps:

Step (1): Initialization: at (l, p)←(0, 0), w←1, after the sub-flows are connected, collect link information l _i , p _i , w _i for each sub-flow.

Step (2): Calculate the theoretical transmission delay of each sub-flow Sort the packet loss rate l _i and theoretical transmission delay d _i of all sub-flows.

Step (3): Scheduling packets according to the theoretical transmission delay of each subflow, for After dX is sorted from small to large, it is d′ _i , the current congestion window size of subflow i is w _i , and the number of confirmed packets is w′ _i .

Step (4): Yes Δw _i ←w _i -w′ _i , according to [d′ _i Δw _i ], dispatch and allocate packets to each sub-flow sequentially.

Step (5): When the sub-flow receives the acknowledgment ack, collect its link information and calculate its theoretical transmission delay at this time. If d _i does not change, proceed to step (4), otherwise, proceed to step (2).

Step (6): When all data transmission is completed, each sub-stream is disconnected sequentially after transmission.

The main idea of the present invention is to dynamically adjust the message of the sub-stream according to the change of the link of the sub-stream, and the sub-stream with a smaller delay sends the pre-order message, so as to avoid the flow control caused by the overflow of the receiving buffer of the better sub-stream It is forced to wait for the poorer substream. During the transmission process, real-time scheduling adjustments must be made whenever the sub-flow link changes. Therefore, MPTCP needs to monitor the link information of each sub-flow in real time. Whenever a sub-flow receives an ack, it must respond to MPTCP. Provides information about the current link.

Above, a kind of multi-path TCP transmission protocol message scheduling method provided by the present invention has been introduced in detail. In this paper, specific examples have been used to illustrate the principle and implementation of the present invention. The description of the above embodiments is only used to help understanding The method of the present invention and its core idea; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation and scope of application. In summary, the content of this specification should not be construed as a limitation of the invention.

Claims (3)

1. a multipath TCP transmission protocol massages dispatching method, is characterized in that, described method comprises:

After setting up subflow connection, the link information of the every bar subflow of real-time collecting; Described link information comprises packet loss, actual transmissions time delay and current congestion window;

Calculate the theoretical transmission time delay of every bar subflow, the packet loss of all subflows and theoretical transmission time delay are sorted;

The scheduling of message is carried out according to the theoretical transmission time delay of every bar subflow, packet loss and congestion window;

The theoretical transmission time delay of the every bar subflow of described basis, packet loss and congestion window carry out the scheduling of message, specifically comprise: A, according to subflow theoretical transmission time delay and packet loss to the sequence of subflow, determine the dispatching message priority of subflow;

B, according to congestion window and confirmation message number determine the message number needed for each subflow;

C, according to the priority of subflow successively to the message number needed for every its current substream of strip flow assignment, meanwhile, the link-quality of each subflow of real-time judgment:

If normal, continue as its distribute needed for message number;

Otherwise, re-execute steps A, B and C.

2. method according to claim 1, is characterized in that:

Described link-quality is judged by subflow theoretical transmission time delay and packet loss.

3. method according to claim 2, is characterized in that:

If packet loss l, actual transmissions time delay p and current congestion window w, then in transmitting procedure, the time that each message success theory can arrive is:

$d=\underset{k=1}{\overset{\&infin;}{\mathrm{\&Sigma;}}}\mathrm{kp}{(1-l)}^k=\frac{(1-l)p}{l^2}$

Wherein, k represents the number of subflow.