Convergence and performance comparison of three different structures of bootstrap blind adaptive algorithm for multisignal co-channel separation

Abstract

The authors extend the previously reported three structures of bootstrap blind adaptive separators to the multisignal channel case. They suggest a recursive weight updating algorithm for the three structures: power-power, correlator-correlator, and power-correlator. The optimum weights for these separators were found analytically in the absence of noise. The signal separation process was shown via simulation by the output learning curve. It was shown that the different bootstrap separators converge to their steady states almost with the same speed for a two or three signal channel. The steady-state interference residues of the three separators are different, lowest for power-power and highest for correlator-correlator. The use of equalization (automatic gain control) at the output of these structures improves the depth of interference cancellation dramatically.