A transport layer protocol for cognitive radio networks

Abstract

For years, TCP has been the first choice for data transportation for the Internet. It provides reliable data delivery with the help of its flow control and congestion control mechanism. In order to improve TCP performance many modifications were proposed to the TCP congestion control mechanisms. However, some of the features of cognitive radio networks make TCP perform poorly in terms of throughput. Some of these features of cognitive radio networks are frequent bandwidth variation, licensed user interruption, and disconnections due to spectrum sensing. These features cause packet losses and time-outs which are mistakenly categorized as congestion losses by TCP. In this thesis, the TCP framework was modified to work efficiently in cognitive radio networks. Markov model that captures the behaviour of TCP is developed and used to evaluate the performance of the proposed protocol.