基于Fuzzy丢包区分的TCP拥塞控制算法

TCP Congestion Control Algorithm Based on Fuzzy Loss Package Differentiating

针对传统传输控制协议(TCP)应用于异构网络的局限性,在研究灰色关联度基础上,分析网络参数,提出一种基于往返延迟抖动积区分丢包的TCP-N算法。根据测得的往返延迟抖动积构建隶属函数,区分无线误码丢包和网络拥塞丢包,并依据隶属度进行相应的拥塞控制。仿真实验结果表明,与传统TCP协议相比,TCP-N算法在异构网络中能够较准确地区分无线误码丢包和网络拥塞丢包,提高带宽利用率和吞吐量。

Aiming at the limitation of the traditional Transmission Control Protocol（TCP） control protocol used in the heterogeneous network, in the use of grey correlation analysis on the basis of network parameters, this paper puts forward a TCP-N protocol based on the round delay jitter product differentiating packet loss. According to the measured round delay jitter product, this algorithm constructs membership function, it differentiates wireless error packet loss and network congestion packet loss according to the membership degree to the corresponding congestion control. Simulation experimental results show that compared with the traditional TCP protocol, TCP-N can more accurately differentiate the wireless error loss from congestion loss in heterogeneous network, improve the TCP throughput and bandwidth utilization, and improves network performance.