IMPROVED PROPORTIONATE AFFINE PROJECTION ALGORITHM FOR ADAPTIVE FEEDBACK CANCELLATION

Abstract

Acoustic feedback is a major problem in
open-fit digital hearing aids, which significantly lowers
the signal quality and limits the achievable maximum
stable gain. Adaptive feedback cancellation (AFC) is a
common and efficient approach, however, it introduces
a biased estimate of the feedback path due to a high
correlation between loudspeaker signal and the incoming
signal, especially when the incoming signal is spectrally
coloured, e.g., speech, music. The prediction error method
(PEM) is well known for reducing this bias, resulting in
significant performance improvement. To further improve
the performance of the conventional PEM we propose to
integrate the improved proportionate affine projection
algorithm (IPAPA) into the PEM. The proposed method,
namely PEM-IPAPA, leverages sparse characteristics of
the feedback path and a fast adaptive filtering technique
to enhance the convergence/tracking rates. A detailed
derivation of the proposed AFC method and its stability
analysis are also considered.We evaluate the performance
of the proposed method with recorded speech and music
as the incoming signals, and with an abrupt change of the
feedback path. Simulation results show that the proposed
method achieves higher convergence/tracking rates while
retaining similar steady-state error and signal quality
compared to the state-of-the-art baselines.