J. Basic. Appl. Sci. Res., 2(7)7175-7180, 2012 © 2012, TextRoad Publication ISSN 2090-4304 Journal of Basic and Applied Scientific Research www.textroad.com *Corresponding Author: R. Barzamini Student Member, IEEE, Amirkabir University of Technology, Tehran, Iran (e-mail: barzamini@yahoo.com). A New Sliding Mode Controller for TCP Congestion Control R. Barzamini 1 Student Member, IEEE, M. Shafiee 2 , Member, IEEE Amirkabir University of Technology,Tehran, Iran ABSTRACT In this paper a new sliding mode controller for congestion problem in TCP networks has been proposed. Congestion occurs during high network loads then this congestion focuses on some aspect of network behavior under high load. In a congestion situation three things accrue in sequence. Congestion control is a set of mechanisms that prevent or reduce such as consumptions. Generally, to solve this problem with a systematic approach using control theory, closed loop data transfer processing structure in computer networks can be considered. Sliding mode controller is robust against modeling uncertainties and disturbances. In sliding mode control states should be reached a predefined surface (sliding surface), in a limited time and remain the same surface over time. Moving on the sliding surface is independent of the uncertainties, so this technique is one method of robust control. After applying controller to system, stability of the system with controller has been proven by Lyapunov stability criteria. Simulation result shows the efficiency of the sliding mode controller in different scenarios. KEYWORDS: Sliding mode control, Congestion control, Robust, Uncertainties, Disturbances Lyapunov. I. INTRODUCTION Congestion collapse was identified as a possible problem as far back as 1984 [1]. It was first observed in the early days of Internet in October 1986, when the NSFnet phase-I backbone dropped three orders of magnitude from its capacity of 32 kbit/s to 40 bit/s, to solve the problem the end nodes started implementing Van Jacobson's congestion control between 1987 and 1988 [2]. Congestion occurs during high network loads then this congestion focuses on some aspect of network behavior under high load. In a congestion situation three things accrue in sequence. Firstly, the queuing delay of the data packets increases then there may be packet losses and, finally, the traffic is dominated by retransmissions, so the data rate decreases. Delay, loss and single-bit or multi-bit explicit signals from routers can be used as feedback for congestion control. Congestion control is a set of mechanisms that prevent or reduce such as consumptions. Generally, to solve this problem with a systematic approach using control theory, closed loop data transfer processing structure in computer networks can be considered. Control theory shows good ways to solve this problem. The word control has a double meaning; first, controlling a system can be understood simply as testing or checking that its behavior is satisfactory. In a deeper sense, to control is also to act, to put things in order to guarantee that the system behaves as desired. Control theory is an interdisciplinary branch of engineering and mathematics, which deals with the behavior of dynamical systems. Control Theory shows systematic approach to analysis and designs a system, predict system response to some input and approaches to assessing system stability. Active Queue Management (AQM) is the most famous congestion controller for TCP networks. While the rest of the components are altogether defined as a plant in figure 1 [3],[7], AQM is designed as a congestion controller to be implemented in network routers. Figure 1: The feedback control system for TCP/AQM. The routers are run on by AQM algorithms and they detect incipient congestion through typically monitoring the instantaneous or average queue size. AQM algorithms infer congestion on the link and notify the end systems to speed down their transmissions by proactively dropping some of the packets arriving at a router or by marking the packets when the average queue size exceeds a certain threshold. End systems that experience the marked or dropped packets reduce their transmission rates for relieving congestion and preventing the queue from overflowing. Congestion is prevented With AQM 7175