On the Distortion Exponent of Partially Closed-Loop Wireless Systems

In wireless multimedia communications, optimizing the quality of user experience is a major design goal, which is often quantified by the end-to-end distortion between the actual multimedia source at the encoder and its reconstructed version at the decoder. Joint source-channel coding (JSCC) techniques aim to optimize codec and radio system parameters in order to minimize end-to-end distortion and yield enhanced quality of user experience when compared with separate source-channel coding (SSCC) techniques. In this paper, we consider partially closed-loop wireless systems, where the channel encoder has full channel state information, but the source encoder knows only the long term channel statistics. Using the distortion exponent as the figure of merit, we determine the optimal JSCC scheme with layered coding for a point- to-point block-fading wireless channel. Furthermore, we investigate the performance of JSCC and layered coding techniques in a realistic WiMAX-based system- level simulation (SLS) environment, and demonstrate their advantages over separate source-channel coding (SSCC) techniques in terms of reduced end-to-end distortion and higher peak signal-to-noise ratio (PSNR).