摘要
无线局域网中,传统拥塞控制算法往往把因信道错误引发的丢包归因于网络拥塞,进而不必要地减小拥塞窗口(cwnd),导致传输性能下降.为了解决这个问题,TCP Westwood(TCPW)算法通过监听ACK流估计网络带宽,以此调整拥塞算法的参数.在TCPW中引入更准确的CAPPROBE带宽估计,提出并实现了一种TCPW CARPROBE的网络拥塞算法(TCPWC).该算法通过设计ACK双包机制和CAPPROBE算法获得更精确的瓶颈链路带宽估计值,合理调整cwnd和慢启动算法的阈值(ssthresh),避免拥塞控制机制的过度反应,从而提高无线网络的传输性能.在NS3仿真环境实验中,该算法与传统的TCP Tahoe、TCP Reno和TCP NewReno机制相比,不仅提高了网络吞吐量,而且具有友好性和公平性,从而验证算法的优越性.
Traditional TCP protocols regard all packet losses as the signals of network congestions in wireless networks, and then immediately decrease Congestion Window （cwnd） which results in the unnecessary degradation of transmission.For purpose of solving this problem,TCP Westwood （TCPW） monitors the stream of acknowledges （ACKs） for obtaining more information to esti- mate the bandwidth of links.The bandwidth estimation is then used to adjust the parameters of congestion control algorithms.We in- troduce a more accurate bandwidth estimation,thus present and implement a new network congestion algorithm TCPW CAPPROBE （TCPWC）.TCPWC acquires more accurate bandwidth by designing ACK pairs based on the CAPPROBE algorithm.Then we adjust cwnd and Slow Start Threshold （ssthresh）, thus avoid overreaction of congestion control mechanism.The simulation results in Network Simulator ns-3 have shown that TCPWC outperforms TCPW and other existing TCP versions such as TCP Tahoe,TCP Reno, and TCP NewReno in terms of network throughput,fairness and friendliness.
出处
《厦门大学学报（自然科学版）》
CAS
CSCD
北大核心
2014年第4期469-476,共8页
Journal of Xiamen University：Natural Science
基金
国家自然科学基金(61271242
61100033
61001072
61379157)
国家重大仪器设备开发专项(2011YQ030124)
中央高校基本科研业务费专项资金(2010121066
2012121028)
福建省自然科学基金(2013J05101
2012J01286)
四川省2013科技计划项目(2013GZ0016)
深圳市战略性新兴产业发展专项资金(JCYJ2012)
