Performance comparison of a distributed source coding scheme using Log-MAP and SOVA decoder in an AWGN channel

This paper presents a distributed source coding scheme for two correlated image data using Log-MAP and SOVA decoders. Both decoders’ peak signal to noise ratio (PSNR) performance has been studied in symmetric as well as asymmetric turbo codes environment. Turbo codes are promising for distributed source coding (DSC) because of their simple encoding implementation and impressive decoding performance. Owing to simple encoding and impressive coding efficiency, DSC is becoming one of the enabling technologies for sensor networks. In this paper, the impact of data correlation and additive noise (AWGN) on distributed source coded image data evaluated in terms of its PSNR performance. This work also evaluates the performance by transmitting the side information through an AWGN channel unlike earlier proposals which assume perfect side information at the decoder. It is observed that Log-MAP decoder performs better than SOVA decoder in the distributed source coding environment also although DSC with Log-MAP decoder is more computationally complex than that of the SOVA decoder. Simulation results are presented for different degrees of correlation between two image data using Log-MAP and SOVA decoding algorithms. It is also observed that a higher correlation corresponds to better side-information and it improves the performance of image reconstruction in both the decoding algorithms.