Improving TCP Performance based on Fuzzy-Logic for Mobile Ad hoc Networks

Transmission Control Protocol (TCP) is designed for wired networks and the sender assumes that packet loss is an indicator of network congestion. However, this assumption may not apply to Mobile Ad hoc Networks (MANETs). Numerous transmission control protocols have been proposed in the literature to address the problem of congestion in MANETs. Their performance is highly dependent on networks conditions, probe and explicit notification messages. This paper proposed a fuzzy logic based inference system based on the factors of expected throughput and actual throughput to dynamically adjust the congestion window size to improve the performance of TCP in MANETs. Fuzzy inference system is an attractive approach used in MANETs which suffer from all kinds of uncertainty, randomness and vagueness. The scheme does not rely on any explicit feedback from the network and requires only sender side modifications. The simulation study examine the throughput, fairness index degree, packet loss percentage and packet delay as performance metrics of the ad hoc network with regard to equal sharing of network bandwidth among multiple TCP flows. The results observe that the proposed scheme achieves the desired goals of performance improvement in terms of packet loss and packet delay compared to TCP-Vegas, Reno and New Reno.