Noncooperative Beamforming for Ad hoc Networks

University Of California-Santa Barbara, Santa Barbara CA

Investigators

Abstract

Current wireless data networks, such as 802.11b/g/a (Wi-Fi) which are based on single antenna transmission are inefficient in terms of the number of users (capacity) that can be supported in a given area. It is well known that multiple antenna elements (arrays) can be combined with beamforming and space-time coding to greatly enhance the capacity of cellular networks. However, the use of antenna arrays to enhance capacity in Ad hoc networks (i.e. when a base station is not available) is not well understood. The development of new beamforming and space-time coding techniques that exploit antenna arrays in Ad hoc networks will greatly increase the number of users and data rates that can be supported in future Wi-Fi and Wi-Max type applications, and reduce dependence on cellular and optical backbone infrastructures. This project specifically develops and analyzes a class of distributed array processing algorithms for increasing Ad hoc wireless network capacity based on iterative minimum mean-square error (IMMSE) beamforming. In IMMSE, the transmit beamformer at each node is set to the conjugate of the conventional MMSE receive beamformer computed using a training sequence. IMMSE is then studied using noncooperative game theory in order to obtain convergence and efficiency results. The IMMSE algorithm is extended to space-time coding to yield approximate solutions to maximizing decoupled Shannon capacities of Ad hoc networks. These IMMSE-Deflation (IMMSE-D) algorithms attempt to balance desired link capacity with a tax on interference to other links, and hence may offer better network throughput than greedy capacity-maximization algorithms. Overall network throughput is evaluated by embedding the IMMSE and IMMSE-D algorithms in a network simulation software package.