Performance of VBR video with equalization on wireless fading channels

Abstract

The error and delay performance of variable bit rate (VBR) video transmission over non-ideal channels is examined using equalization schemes. The temporal correlation features of the fading signal and the source video are modeled using finite state Markov chains. The source and training signals are multiplexed using a time slotted multiple access scheme. For slow-fading channels, the characteristic time-scale of the fade durations, the step-size parameter of the LMS equalizer training algorithm, and the number of training sequences allocated per time slot are found to jointly influence error performance. Simulation results show that the proposed channel access scheme can lead to an order of magnitude decrease in the bit error ratio if the combination of training overhead and step-size parameter are chosen judiciously with respect to the fading time-scale. In addition, the multiplexing efficiency of VBR encoded video is shown to be useful for limiting the channel access delays when the training overhead is increased.