The MVDR Beamformer Based on Hypercomplex Processes

In the paper, the problem of MVDR beamformer based on hypercomplex processes is investigated in a scenarios where there exist one signal and one interference that are uncorrelated. First, a quaternion model of linear array with two-components EM vector-sensors is presented. Based on the quaternion model, a quaternion MVDR (QMVDR) beamformer is derived. The quaternion-valued output y(n) of the QMVDR consists of two complex components y1(n) and y2(n). In y2(n), there exist only the interference and noise components, but no desired signal. The desired signal is included in y1(n) and it is corrupted by the interference and noise. To extract the desired signal, we can employ y2(n) to cancel partly the interference component in y1(n). Thus, an interference and noise cancellation (INC) algorithm of QMVDR is proposed. The INC algorithm of QMVDR beamformer is similar to the generalized sidelobe canceller. Simulation results show that the proposed algorithm can achieve a much better performance in terms of output SINR than existing ones.