Performance of transmission control protocol on self-similar traffic.

Abstract

Internet traffic will continue to expand in terms of both volume and users, and transmission congestion protocol (TCP) is accounting for more than 90% of the total traffic volume. It is well known that network traffic exhibit self-similar behavior and cannot be described by traditional Markovian models such as the Poisson process. The burstiness that occurs in self-similar traffic needs to be managed so as to enable optimal planning and to achieve higher quality of service. In relation to this, there must be a full understanding of the global network traffic especially on the impact of self-similar traffic on network protocol. In this paper we focus on experimental quantitative comparison of four TCP variants namely Tahoe, NewReno, Vegas, and SACK running over self-similar traffic. With the aid of network simulator, NS2, the impact of TCP microflow is observed, and data analysis such as packets dropped and retransmission, throughput, and behavior of the data received over time is analyzed. We employed optimization method to measure Hurst parameter, the degree of self-similarity. Our study shows that at low and highly congested traffic, network traffic does not exhibit self-similarity. In addition TCP microflow exhibit self-similar behavior when traffic load is moderately congested and our result revealed that Vegas are the best performance TCP variant in self-similar traffic.