Abstract

Forward Error Correction (FEC) coding has been widely used in digital communications to improve the transmission performance over noisy channels. Based on the introduced redundancy and encoding structure, a maximum likelihood (ML) decoding scheme can be used to correct significant part of transmission errors. For digital video transmission, further enhancement of the reconstructed video signal can be done by error concealment based on the residual redundancy of the video signal. FEC and error concealment are usually treated as independent processes. This thesis presents an error resilience scheme, which combines Forward Error Correction (FEC) and Error Concealment (EC) in order to further enhance the quality of the video signal transmitted over a noisy channel. Of particular, for a given FEC scheme in use, a Multiple Candidate Likelihood (MCL) channel decoding strategy is proposed. Unlike the traditional ML decoder that provides only one corrected sequence, the proposed MCL decoding strategy offers a number of candidates with their reliability information. These candidates are further examined in terms of syntax/semantic validity and their discontinuity measure, which are related to the video source coding. Performance of video signals in terms of the peak signal-to-noise ratio (PSNR) as well as subjective measure by visual inspection of video frames and sequence is investigated for different video sequences and various FEC coding schemes. Both simulation and analysis are used to evaluate performance of the proposed schemes and to assess their complexity. Research results indicate that the proposed strategy outperforms the traditional ML decoding technique for the same FEC code in use, especially at the high channel bit error rates (BER). Simulation results are in a good agreement with the analytical prediction. Selection of key parameters and characteristics of the proposed combined FEC and EC are discussed.