A method for determining beamforming weights used onboard a satellite and ground-based beamforming weights used in a ground-based station as part of a satellite communication system. This beamforming method is a two-stage beamforming process that requires a reduced downlink bandwidth between the satellite and the ground-based station yet achieves optimal signal-to-noise ratio for bandwidth allocated for the downlink. values for the fixed onboard beamforming weights are computed to yield a maximum, max A ⁢ ( min U ⁢ S N ⁢ | W = W ⋓ ) , where the maximum is computed over all possible fixed weights A represented by an L×M matrix, the minimum is computed over all possible positions of remote communication devices U, and the signal-to-noise ratio (S/N) is computed for the optimal set of ground-based beamforming weights W={hacek over (W)}.