A set of multiple training signals to be transmitted via multiple antennas over a communication channel are generated, and a different antenna weight vector (AWV) from a first set of AWVs is applied to each of the multiple training signals as the training signals are transmitted during a current iteration of a beamforming procedure. Feedback generated using the multiple training signals is received. A second set of AWVs to be used in a next iteration of the beamforming procedure is determined using the feedback.