Energy efficient weighted sampling matrix based CS technique for WSN

Sensor nodes have energy constraints which significantly reduces their lifetime. Compressed Sensing (CS) can be used to achieve compression and energy efficiency, thereby increasing the lifetime of WSN. However, signal components are projected in a random direction which produces poor performance on complex signals. In order to achieve efficiency, adaptive projection of signal components is necessary. Hence, block CS (BCS) with Haar DWT (HDWT) based image representation and Energy based Re-weighted Sampling (ERWS) techniques are proposed to extract the high energy components of the image. Analysis confirms that the proposed method outperforms conventional techniques even for fewer measurements and reconstructed image quality is greatly enhanced. Experimental analysis is done using Atmega128 of Mica2 mote. Execution time and energy consumed are computed in the hardware platform. HDWT based ERWS has approximately 34.5 % lesser energy consumption than DCT based BCS techniques.