Research And Optimization Of Congestion Control Algorithm Based On TCP BBR

In today's society,a small smart phone runs a variety of businesses related to daily life,and the smooth operation of each business is inseparable from the technical support of network communication.As the cornerstone of communication protocol in network communication,the improvement of TCP can improve the efficiency of people's life and work.However,the development of wireless transmission technology has changed the communication environment dramatically.In modern communication network,the communication bandwidth is getting larger and larger,the transmission delay is getting longer and longer,and there are more and more interference in the transmission process.As a result,packet loss occurs not only in the router queuing process,but also in the wireless transmission process,which seriously affects the communication performance of TCP.Therefore,the study of TCP congestion control algorithm has also become the focus of scholars.In this paper,two default congestion control protocols in Linux system,TCP New Reno,TCP Cubic and the latest TCP BBR algorithm proposed by Google,are studied.In this paper,test scenarios are built on PC and mobile terminals respectively,and TC and iperf tools in Linux system are used to set channel parameters and monitor test results.The channel parameters include the round-trip delay and packet loss rate of the link,while the performance comparison focuses on the index of throughput rate.This paper draws a number of curves according to the test results,analyzes the reasons for the performance differences of the three algorithms according to the curves,and focuses on the differences between BBR algorithm and other algorithms.After the test results are obtained,this paper proposes a solution to the problems existing in the current BBR algorithm.For example,the sensitivity of BBR algorithm to bandwidth is not high,and the scenario of wireless network delay jitter has certainbandwidth loss.Because the BBR algorithm proposed late,on the BBR algorithm research data is not much,so this article in-depth source code,from the BBR algorithm principle and mathematical perspective,detailed analysis of the above problems,and put forward an optimization scheme.In order to test the feasibility of the scheme,this paper implemented the optimization scheme in the Android kernel,and compiled and passed it.With the help of speedtest software,a lot of tests were conducted to compare the standard BBR algorithm in the real environment such as railway,laboratory,etc.,and the results showed that the optimization scheme was improved to a certain extent compared with the standard BBR.In addition,this paper also introduces the steps of building test scenarios in mobile terminal in detail to help later researchers spend more time on BBR algorithm itself.