Transmit Beamforming for Optimizing Network Capacity

Optimal transmit beamforming is obtained by minimizing the total network transmit power subject to a nodelevel channel capacity constraint. The algorithm that achieves this is referred to as locally enabled global optimization (LEGO)1 and is especially efficient for networks with reciprocal internode transmit and receive channel responses, typically realizable in time division duplex (TDD) networks. The algorithm achieves its network wide performance boost using only local information available at each transceiver. The LEGO algorithm combines the features of multichannel diversity combining, reminiscent of information theoretic water filling arguments, with the ability to exploit MIMO channel and internode transmit diversity, even in the presence of network-wide co-channel interference. A convergence proof is provided for a variant of LEGO and a series of numerical experiments are shown, which reveal an order of magnitude capacity improvement over conventional single antenna systems, and a several order of magnitude improvement in required transmission power.