Performance of blind beamforming algorithms

Many different methods for beamforming using partial or no a priori information about sensor locations and response patterns have been proposed in recent years. This paper investigates the performance of several of these methods for a variety of signal and interference types, sensor geometries, multipath, and sensor mismatches. The methods investigated include Cardoso and Souloumiac's (1993) JADE algorithm, Laheld and Cardoso's (1994) EASI method, van der Veen and Paulraj's (1996) ACMA, and Gonen and Mendel's (1997) fourth-order ESPRIT-like method. The signal types investigated include CW signals, QAM-4 signals, and wideband signals. These are simulated for situations with linear arrays and random arrays. The simulations indicate that the JADE algorithm works well in a wide variety of situations.