Abstract—Due to the difficulty in discriminating the wireless loss from congestion loss in heterogeneous networks, TCP Westwood can not accurately estimate the available bandwidth, and cause low network utilization. In this paper, we propose an improved congestion control algorithm called TCP NewBR, which adaptively estimates bandwidth by using the bottleneck link utilization, and selects more precise time interval of ACK to improve the accuracy of available bandwidth estimation. In addition, we also modify the faster recovery and faster retransmission algorithms according to the queue length of bottleneck link. Simulation results in several scenarios show that TCP NewBR can get more accurate bandwidth estimation, and more significant throughput than traditional TCP Westwood, and fairer and friendlier than TCP Reno.

Index Terms—Heterogeneous networks, bandwidth estimation, throughput, TCP Westwood

Cite: Rui-Qing Wu, Hong Jie, and Nan Ding, "Improved Multi-Channel Blind De-Convolution Algorithm for Linear Convolved Mixture DS/CDMA Signals," Journal of Communications, vol. 9, no. 10, pp. 792-797, 2014. Doi: 10.12720/jcm.9.10.792-797