Fast-AIMD: A fairness based congestion control approach for TCP networks

The majority of real time applications require hasty data transmission over a TCP channel in the network. It is a necessity of fair distribution of resources to enhance the performance of the network. Additive increase and multiplicative decrease (AIMD) is one of the finest congestion control technique for TCP networks, achieves a responsibly good fairness and throughput, which reduces packet loss rate. This paper proposes an AIMD based congestion control technique to deal with the fairness and attain the better and fast bandwidth utilization. We allow our mechanism to increase the congestion window size exponentially by a multiplicative factor at slow start and drop off the congestion window with an additional increase for every packet loss. Our approach uses the concept of pre-determination of congestion window size for each multiplicative increase with respect to the available bandwidth of the channel. We called our approach as Fast-AIMD (FAIMD). The simulation results demonstrate that our mechanism can accomplish a faster growth in congestion window of each flow as compared to the existing mechanism for TCP networks.