A computationally efficient distributed beamforming technique for multi-user relay networks is developed. The channel state information is assumed to be known at the relays or destinations, and the total relay transmitted power is minimized subject to the destination quality-of-service constraints. It is shown that this problem can be approximately converted to a convex second-order cone programming form. As a result, the proposed network beamforming technique offers a substantially reduced computational complexity than earlier state-ofthe-art techniques that are based on semidefinite relaxation.
Haihua Chen, Alex B. Gershman, Shahram Shahbazpana