On Block Compressed Sensing far end reconstruction using OFDM

Abstract

Compressed Sensing (CS) allows images to be reconstructed from fewer measurements compared to that proposed by Nyquist. To reduce the computational expense, instead of applying CS to the entire image, it is judicious if CS were applied in blocks. Sensing an image block by block not only requires lesser memory but also eases the computational burden during reconstruction. Although a large amount of work has been done in integrating CS and OFDM, empirical analysis of BCS in conjugation with OFDM has not been done so far. This work senses an image using Block Compressed Sensing (BCS), encodes it using DPCM and transmits it in the OFDM framework Among the reconstruction algorithms OMP and SPL, this paper attempts to come up with the superior algorithm after performing a detailed comparison by varying block sizes, sub-sampling rates per block and the AWGN channel SNR. The necessity of the OFDM scheme for practical deployment has also been re-asserted strongly.