Compressed sensing of audio signals using multiple sensors

Compressed sensing is an attractive compression scheme due to its universality and lack of complexity on the sensor side. In this paper we present a study on compressed sensing of real, non-sparse, audio signals. We investigate the performance of different bases and reconstruction algorithms. We then explore the performance of multi-sensor compressed sensing of audio signals and present a novel scheme to provide improved performance over standard reconstruction algorithms. We then present simulations and measured results of a new algorithm to perform efficient detection and estimation in a sensor network that is used to track the location of a subject wearing a tracking device, which periodically transmits a very sparse audio signal. We show that our algorithm can dramatically reduce the number of transmissions in such a sensor network.