Jointly optimal congestion control and power control in wireless multihop networks

Abstract

Power control can significantly enhance the performance of congestion control mechanisms, such as TCP, in wireless networks. We present a distributed power control algorithm that works together with the original TCP protocol to increase end- to-end throughput and energy efficiency of multihop data transmissions in CDMA wireless ad hoc networks. We prove that the resulted nonlinear coupled system converges to the global optimality of network utility maximization with elastic link capacities. This cross-layer algorithm can be interpreted as using link queuing delays as shadow prices to coordinate bandwidth demand and supply. Various simulations show desirable properties of the algorithms, including robustness to channel variations and fading estimation errors, and flexibility in the tradeoff between performance optimally and algorithmic simplicity.