Iterative decoding of FEC based multiple description codes for image transmission over wireless channel

Multiple description coding has emerged as an attractive framework for robust transmission of multimedia information over packet loss and fading wireless channels. Forward error correction (FEC) based multiple description coding enables generation of an arbitrary number of descriptions from a progressive image bitstream. These descriptions are formed using erasure protection properties of Reed-Solomon codes and each code is spread across multiple packets. The level of protection is determined based on the relative importance of the information symbols in the progressive bitstream. Each description is then encoded using concatenation of convolutional code and CRC. This paper proposes an iterative soft decision decoding by exchanging extrinsic information between Reed-Solomon code across packets and convolutional coded descriptions transmitted over wireless channels, in contrast to existing approaches where hard decision is performed on the outcome of the inner channel code and outer FEC (Reed-Solomon code) is used to fill-up the erased descriptions. The simulation results over Rayleigh flat fading channel show significant improvement in received PSNR compared to baseline system.