Simulation based performance evaluation of TCP variants along with UDP flow analysis of throughput with respect to delay, buffer size and time

摘要

In a data communication network, several flows contend to utilize limited and shared network resources, however, this gives rise to congestion along with other network issues. So, this leads to the development of congestion control based methods/ protocols to support the deployment of real-time multimedia applications while ensuring reliability and fairness. This paper presents a simulation based study, analyzing variants of TCP i.e. TCP Tahoe, TCP Reno and TCP New Reno. In this work a simulator, network simulator-3, NS-3.25, has been used to analyze the throughput for different TCP flows in presence of a UDP flow running under a bad-sensitive and lossy environment. Further, performance is evaluated on the basis of simulation results obtained under a realistic scenario built in NS-3, emulating different background traffics, link delays, bottleneck link, limited queue size, and with different link bandwidths. The topology laid is simulated for two different queue management techniques i.e. Drop Tail (DT) and Random Early Drop (RED), operating while utilizing different congestion avoidance capabilities. Goal is to evaluate protocols in stiff conditions while evaluating each protocol based on different performance metrics. Observed experimental results have shown that TCP New Reno is more robust and has given persistent performance levels in terms of throughput in a constrained environment.