Robust Power Allocation for MIMO Beamforming under Time Varying Channel Conditions

We consider the downlink transmit power allocation problems in multi-user MIMO wireless networks using zeroforcing beamforming. Traditionally such problems are solved by water-filling algorithm under the assumption of perfect channel knowledge. However when channel information is not known a priori or time varying the water-filling solution is shown to be unstable. We use the sliding mode control theory to synthesize the transmit powers so that the target SINR requirements of all users are met. We synthesize the sliding mode controller for the case of zero-forcing beamforming. The synthesis problem is solved under time varying Rayleigh fading channel conditions. Our solutions and simulation results show that our sliding mode controller is stable and delivers better quality of service under practical channel conditions.