A tcp congestion control method based on sdn network

Abstract

The invention discloses a TCP congestion control method based on an SDN network. The method comprises the steps: enabling an SDN controller to collect real-time network time delay, time delay jitter, usable network bandwidth, network congestion conditions and information whether a target is reachable or not, and transmitting the above information to to-be-adjusted TCP data flows; and respectively adjusting the to-be-adjusted TCP data flows according to the corresponding adjustment targets. The method can quickly sense the network time delay, congestion condition, bandwidth and error codes through interaction with the SDN controller. Especially for a high-time-delay and high-error-rate wireless network environment, the method can discriminate air interface package loss and network congestion package loss quickly and accurately, enables a TCP protocol to be able to quickly sense the network condition, and quickly adjusts the congestion control mechanism of the TCP protocol, thereby improving the transmission efficiency of the TCP protocol.

Description

technical field [0001] The invention relates to the research field of network congestion, in particular to a TCP congestion control method based on an SDN network. Background technique [0002] In recent years, the explosive growth of the Internet has led to the emergence of a large number of various network applications, the amount of data on the network is increasing, and the load on the network is getting heavier, which leads to congestion. TCP is currently the most widely used transmission protocol in the Internet, and the congestion control algorithm used in TCP is an important factor to ensure the stability of the Internet. TCP runs on an end-to-end virtual connection, mainly based on congestion control to provide reliable datagram transmission in unreliable networks. [0003] TCP was originally designed for wired networks, where network congestion is considered to be the main cause of packet loss and retransmission timer timeout (RTO, Retransmission TimeOut). In addi...

AI Technical Summary

Problems solved by technology

[0004] (1) TCP judges congestion through the repeated confirmation of data segments and the RTP timeout mechanism. In wireless networks, most packet loss problems are not caused by network congestion, so TCP is prone to misjudgment and misresponse problems. ;

[0005] (2) After TCP is started, the source end increases the congestion window every time it receives a new ACK to send as much data as possible. However, the delay and delay jitter in the wireless network are large, resulting in the extremely high timing of ACK arrival. Irregularity, which leads to the instability of the window growth process, that is, the problem of too fast or too slow growth

Images