Marc Moonen

MMoonenMarc Moonen (Electrical Engineering Department, KU Leuven, Belgium)
Distributed adaptive node-specific signal estimation in wireless acoustic sensor networks

Wireless sensor networks (WSNs) have attracted tremendous attention in recent years and have initiated a paradigm shift in the way sensor signals can be acquired and analyzed. The common setup for WSNs is to have many densely deployed inexpensive sensor nodes and to rely on distributed sensing and computation as well as communication to perform an overall network-wide task. The advantage of WSNs compared to traditional fixed sensor arrays (such as a microphone array on a HA) is that many sensors can be used that cover the full spatial field. The challenge, however, is the fact that the sensor positions are usually not known a priori, and that all the signals must somehow be combined and processed. This data fusion can happen in a centralized fashion, where all the nodes send their signals to a master node, or in a distributed fashion, where e.g. nodes only share signals with neighbor nodes and the processing is shared between the nodes. WSNs have become popular for parameter estimation tasks, e.g., for the estimation of temperature, chemical composition, wind speed, etc., where parameters vary slowly over time and so low sampling rates can be used. The estimates can then be iteratively refined by information diffusion over the network. However, it is difficult to apply such iterative techniques in signal estimation tasks in general and applications with audio signals in particular, due to the high sampling rates and the rapidly changing characteristics of typical audio signals (e.g. speech signals).

The talk will focus on recently developed so-called distributed adaptive node-specific signal estimation (DANSE) procedures in WSNs with a fully connected topology or a non-fully connected topology such as a tree topology. The signal estimation is fully distributed and node-specific, meaning that each node performs its own signal estimation task. The algorithm applies to noise reduction scenarios in wireless acoustic sensor networks where the nodes have multiple sensors/microphones, such as binaural hearing aids and in hearing aid networks. A strategy is provided to reduce the number of signals that each node transmits to other nodes. Instead of transmitting all its sensor signals, each node then transmits only one or a few optimal linear combinations of its sensor signals and possibly also of the signals that the nodes receives from other nodes, in a non-fully connected topology. In general, this provides a significant communication bandwidth and computational complexity reduction. Nevertheless, for specific scenarios (satisfying a specific model assumption) optimal signal estimation is still achieved in each node as if all sensor signals had been transmitted to all nodes. In speech enhancement scenarios the specific model assumption has indeed been shown to hold. The talk will focus on theoretical background and algorithms as well as on practical applications an implementation aspects.

Marc Moonen received the electrical engineering degree and the PhD degree in applied sciences from KU Leuven, Belgium, in 1986 and 1990. Since 2004 he is a Full Professor at the Electrical Engineering Department of KU Leuven, where he is heading a research team working in the area of numerical algorithms and signal processing for digital communications, wireless communications, DSL and audio signal processing. He is a Fellow of the IEEE (2007). He received the 1994 KU Leuven Research Council Award, the 1997 Alcatel Bell (Belgium) Award (with Piet Vandaele), the 2004 Alcatel Bell (Belgium) Award (with Raphael Cendrillon), and was a 1997 Laureate of the Belgium Royal Academy of Science. He received a journal best paper award from the IEEE Transactions on Signal Processing (with Geert Leus) and from Elsevier Signal Processing (with Simon Doclo). He was chairman of the IEEE Benelux Signal Processing Chapter (1998-2002), and President of EURASIP, the European Association for Signal Processing (2007-2008 and 2011-2012). He has served as Editor-in-Chief for the EURASIP Journal on Applied Signal Processing (2003-2005), and has been a member of the editorial board of IEEE Transactions on Circuits and Systems II, IEEE Signal Processing Magazine, Integration-the VLSI Journal, EURASIP Journal on Wireless Communications and Networking, and Signal Processing. He is currently a member of the editorial board of EURASIP Journal on Applied Signal Processing and Area Editor for Feature Articles in IEEE Signal Processing Magazine. He has supervised 33 finished PhD projects, and has (co-)authored over 200 peer-reviewed journal papers.

Other keynotes