ITR/SI(CISE):Optimal and Robust TCP Congestion Control

California Institute Of Technology, Pasadena CA

Investigators

Abstract

Internet is undergoing an overhaul unprecedented in size, diversity, and reach, with profound im- pact in all aspects of our scientific, social, economic and political life through the integration of networks of communication, transportation, entertainment, utilities, and finance. The stability and robustness of this vital infrastructure demands a rigorous theory to understand the current protocols and evolve them to meet emerging challenges. We propose to develop such a theory for TCP congestion control, and use it to drastically improve the stability, robustness and optimality of the current protocols. A key insight is to view congestion control as a distributed asynchronous computation to maxi- mize aggregate source utility over the Internet; different TCP and active queue management (AQM) schemes correspond to different utility functions and different algorithms to maximize them. Our research hastwo components. First, we will develop a new theoretical model of TCP congestion control based on duality in optimization and multivariate robust control. The theory will clarify the role of source algorithms, such as Tahoe, Reno and Vegas, and active queue management, such as DropTail, RED and REM, in the control of networks and establish performance limits of the current protocols; it will explain the effect on stability when delay, topology, capacity, and load scale up; and it will provide conditions under which the feedback stability ofTCP/AQM algorithms are invariant to these effects. Indeed, such a theory is already emerging from our recent works. Even in its currently preliminary stage, it already provides a fundamental understanding on some widely observed performance and fairness behavior of the current protocols, and uncovers new and surprising stability problems. For example, it shows that the current protocols become unstable and exhibit bifurcation when network capacity increases. Moreover, maintaining stability as capacity scales up arbitrarily imposes severe constraints on how sources adjust their rates (TCP) and what congestion information is fed back (AQM). The current protocol does not satisfy the condition for such stability invariance, and hence may be ill suited for future networks where, pulled by application demand and pushed by technological advances, the capacity will be large. The second component of our research is the design of practical TCP and AQM protocols based on the theory, and the development of prototypes and experiments to demonstrate their effectiveness. We will use the theory to identify the sources of instability in the current protocols when delay, network size, capacity, and traffic load scale up. We will design both enhancements that incrementally evolve the current protocols, and drastically new protocols that have the strongly robust stability property promised by theory. As a concrete application of our algorithms, we will apply them to improve TCP performance over wireless links, both because they are ubiquitous and because they are likely to remain the most important bottlenecks in future networks.