Performance analysis and optimization of progressive image transmission in memory channels

We proposed a new approach for evaluating the performances of the embedded image streams over finite state Markovian chain (FSMC) channels. Based on the fact that the first error-free run model holds for any noisy channels, we obtain its average length in closed-form by error sequence analysis (ESA). The method applies to any valid FSMC channels. It is further shown that some of the existing performance evaluation methods are special eases of this new approach. We then use the results to optimize the performance of equal error protection (EEP) schemes in FSMC channels and investigate the effects of mobile speed on system performance. It is discovered in the simulation that statistical averaging (SA) approach exaggerate the PSNR performance improvement for up to 5 dB. The PSNR performance improvements of the new approach over the single worst BSC based method can be up to 3 dB.