Cross Layer Optimization Of TCP Transmission In Wireless Networks

TCP(Transmission Control Protocol)is the most important and widely used transport layer protocol in communication transmission.In different network environments,the choice of TCP congestion control algorithm has a great influence on end to end transmission service.TCP is a connection oriented and reliable transport layer protocol.The main reason for packet loss in the wired network is network congestion.However,in the wireless network has many problems,such as limited bandwidth,high error rate,high latency,connection interruption,etc.,the traditional transmission protocol is difficult to adapt to the rapidly changing wireless network environment,resulting in low throughput and high latency problems.This paper mainly studies the existing problems of TCP in wireless network,and establishes MPP/G/1 model of data link layer and transport layer based on queuing theory.In order to improve the performance of wireless network TCP,a cross layer optimization based on MMPP/G/1 modelis presented.First,the WDL(Westwood LTE)algorithm is proposed,which improves the bandwidth estimation and congestion control part of the Westwood algorithm,and verifies that TCP can not distinguish the packet loss type in the wireless network.Through LTE simulation platform,the performance of WDL,NewReno and Westwood is compared and analyzed.In addition,the cross layer simulation model of MMPP/G/1 based on queuing theory is built in the wireless network,the steady state distribution,the inter arrival time distribution,the probability distribution of the queue length and the probability distribution of virtual waiting time can be obtained in the system,according to the current status of the queue can calculate the packet loss and throughput in wireless network.The MMPP/G/1 model of 2 states is implemented by simulation,and the correctness of the theoretical model is verified.Based on the previous research,a cross layer algorithm of QL-WT(Queue Length and Wait Time Detection)is proposed.The layer and layer information interaction between the transport layer and the data link layer can more quickly and accurately select the most appropriate transmission rate.By establishing a feedback mechanism model,the queue length and waiting time are fed back to the sending side of TCP and the adaptive change of TCP sending end congestion window is achieved.In simulation,NewReno algorithm and QL-WT algorithm are implemented by Python.The results show that QL-WT algorithm reduces the packet loss rate of wireless network,and improves the utilization of throughput and bandwidth.