A New Stochastic Differential Equation Model for TCPW RE Window Size

A.-K. Aboul-Hassan, O.A. Alim, A.A. Razek, and M.A.M. Rashad (Egypt)

Keywords

– Windowsize, TCP, TCPW RE, PCSDE.

Abstract

— In this paper, we use Poisson Driven Stochastic Differential Equations (PCSDE) to model the window size behavior of a new version of the Transmission Control Protocol (TCP) running dominantly over computer networks which is TCP Westwood Rate Estimation (TCPW RE). The model considers the Additive Increase Adaptive Decrease (AIAD) congestion control scheme relevant to the transmission protocol. The model performance is compared to its opponent, the Additive Increase Multiplicative Decrease (AIMD) control scheme of the legacy protocol (TCP Reno), under different network loading conditions represented by different loss probabilities and Round Trip Times (RTT). The quality measure we use to carry out the performance comparison in such stochastic network is the achieved average throughput that is determined by the steady state mean window size. A significant increase in mean window size is realized in the case of Westwood RE more than that of the Reno protocol along with all different testing conditions verifying the practical fact of the superiority of the adaptive Westwood protocol to the legacy Reno protocol considering error free links.

Important Links:



Go Back