Mismatched source transition probability matrix for adaptive MAP decoding in MPEG-4 imagery wireless transmission systems

Abstract

Here, we design the efficient technique of transmitting the source transition probability matrix (STPM) by accompanied with the lowest frequency subband (LFS). The entries of this stochastic matrix are adaptively computed by applying the first order Markov model with MPEG-4 zerotree sequences. Because of channel disturbances, we obtain the mismatched STPM at the ML-Viterbi receiver and then employ it for newly computed branch metrics at the MAP source-controlled channel decoder. For analysis, we also evaluate the residual redundancies for both the "Lena" and the "Barbara" images. The system performance is summarized in term of both PSNR (dB) and WER for three types of slow flat Rician fading channels. In the mismatched STPM simulation results, we still obtain the most PSNR improvement of about 0.14 dB.