NCP: Finishing Flows Even More Quickly

The transmission control protocol (TCP) is the major transport layer protocol in the Internet today. TCP and its variants have the drawback of not knowing the explicit rate share of flows at bottleneck links. The Rate Control Protocol (RCP) is a major clean slate congestion control protocol which has been recently proposed to address these drawbacks. RCP tries to get explicit knowledge of flow shares at bottleneck links. However, RCP under or over estimates the number of active flows which it needs to obtain the flow fair rate share. This causes under or over utilization of bottleneck link capacity. This in turn can result in very high queue length and packet drops which translate into a high average file completion time (AFCT). In this paper we present the design and analysis of a Network congestion Control Protocol (NCP). NCP can give flows their fair share rates and hence resulting in the minimum AFCT. Unlike RCP, NCP can also use accurate formula to calculate the number of flows sharing a network link. This enables NCP to assign fair share rates to flows without over or under-utilization of bottleneck link capacities. Simulation results confirm the design goals of NCP in achieving minimum AFCT when compared with RCP.