Multicast transmit beamforming using a randomize-in-time strategy

Recently there has been much interest in using transmit beamforming to provide multi-antenna physical-layer multicasting. A state of the art in this context is the semidefinite relaxation (SDR) method for handling the design optimization. This paper proposes an alternative multicast transmit strategy where the transmit beams are random and time-varying. This proposed strategy consists of two key ingredients: i) the randomizations are guided by SDR, but without the need of rank-one approximation as in the existing fixed transmit beamformer; and ii) channel coding is employed to deal with the time randomizations. We adopt an achievable rate perspective, and show that the gaps between the rates of the proposed stochastic beamformers and the optimal multicast capacity are no greater than 0.8314 bps/Hz, irrespective of any factor such as the number of users served. Our simulation results demonstrate that under a rate-1/3 turbo code setting, the stochastic beamformers can yield an SNR gain of more than 4dB compared to the fixed beamformer, given that the bit error rate is 10e−5.