Adaptive envelope-constrained filtering algorithms for noisy input signals

This paper studies the effect of noise at the input channel on the recursive algorithms recently proposed for designing adaptive filters subject to output envelope constraints. It is found that such an adaptive filter may suffer serious performance degradation due to the use of noisy input signal measurements in the filter update equation. To reduce the noise effect, it is proposed to smooth the noisy input signal by averaging over several recent input signal measurements. Numerical results show that a desirable performance improvement can be achieved using such a moving average window.