Blind source extraction for a combined fixed and wireless sensor network

Abstract

The emergence of wireless microphones in everyday life creates opportunities to exploit spatial diversity when using fixed microphone arrays combined with these wireless microphones. Traditional array signal processing (ASP) techniques are not suitable for such a scenario since the locations of the wireless sensors are unknown and probably vary over time. In this paper we investigate the use of blind source extraction (BSE) techniques in such a combined acoustic sensor network to perform speech enhancement. We present strategies that apply traditional ASP techniques to the fixed microphone array, while simultaneously the spatial diversity provided by the wireless microphones is exploited. Our conclusion is that BSE techniques can be used in a combined wireless and fixed microphone network to perform speech enhancement.