Joint Source-Channel Rate Allocation for Wireless Video Transmission Based on Distortion Estimation

Abstract

Joint source channel coding provides an efficient framework in minimizing end-to-end distortion. In this paper, we propose a method of dynamically allocating the available bit rate between video encoder and channel encoder. The rate allocation algorithm is dependent on the estimated quantization and transmission distortion. We propose a quantization distortion model that is based on residual information and quantization parameter. Transmission distortion is estimated based on error propagation and error concealment distortion. Results show good estimate of the actual distortion and able to estimate the distortion before encoding the frame. The proposed distortion models are used to implement a joint source-channel video coding scheme using standard H.264/AVC encoder. The proposed scheme provides significant improvement in decoded video quality which can adapt to varying channel condition by proper allocation of available bit rate between video and channel encoder. The proposed distortion model can be extended in cross-layer optimization between channel encoder code rate and intra-refresh rate selection.