Publication

Abstract : Recent advancement in the field of wireless sensor networks (WSNs) has enabled its use in a variety of multimedia applications where the data to be handled are large that require more memory for storage and high bandwidth for transmission. As WSNs have limited capabilities in terms of computation, memory, energy and bandwidth, compression becomes necessary. The traditional compression methods consume more energy as well as memory which can be overcome by compressive sensing (CS) technique. CS is an emerging technique for efficiently acquiring and reconstructing the signal by processing the reduced number of samples specified by the Nyquist criterion. The objective of this paper is to implement CS for images using the proposed sensing matrix derived from the Toeplitz matrix and its variants. For reconstruction purpose, an existing greedy orthogonal matching pursuit algorithm is used. The measurements obtained from the framework are transmitted in real time using TelosB nodes under Contiki OS platform. The simulation results are compared with the experimental results, and the performance of the CS framework is evaluated in terms of peak signal-to-noise ratio, storage overhead, energy computation, computational time, transmission energy and end-to-end transmission latency. The results show that the performance of the proposed sensing matrix is better in terms of memory requirement, energy computation and computational complexity when compared with an existing Gaussian matrix.