Compressive Sensing Optimization for Signal Ensembles in WSNs

Compressive sensing (CS) is a new approach to simultaneous sensing and compressing that is highly promising for fully distributed compression in wireless sensor networks (WSNs). While a wide investigation has been performed about theory and practice of CS for individual signals, real and practical cases, in general, involve multiple signals, extending the problem of compression from 1-D single-sensor to 2-D multiple-sensors data. In this paper the two most prominent frameworks on sparsity and compressibility of multidimensional signals and signal ensembles, Distributed compressed sensing (DCS) and Kronecker compressive sensing (KCS), are investigated. In this paper we compare these two frameworks against a common set of artificial signals properly built to embody the main characteristics of natural signals. We further investigate how, in a real deployment, DCS can be used to reduce the power consumption and to prolong lifetime. In particular an extensive analysis is performed using real commercial off-the-shelf (COTS) hardware evaluating how different kind of compression matrices can affect the jointly reconstruction, trying to achieve the better tradeoff between quality and energy expenditure.