Cross-layer Optimized Coding Mode Selection for Wireless Video Communications

Abstract

This paper proposed a coding mode selection method for video transmission over wireless networks. Unlike previous mode selection methods disregarding channel distortion or assuming constant packet loss rate (PLR), this method includes a cross-layer controller to collect both source and channel information. The mode selection process is formulated as a delay constrained distortion minimization problem. The three components in the resulting Lagrange cost function, namely distortion, Lagrange multiplier and packet delay, are estimated with online channel information feedback. Suboptimal coding decision and physical layer modulation and coding scheme (MCS) are determined by the controller for each packet. In our experiment, three coding modes, intra, inter and down sampling, are tested under various channel conditions. Compared to conventional method, 3.6dB to 7.5dB average reduction in distortion is achieved under different channel condition, while down sampling further gains up to 2.2dB distortion reduction in low data rate transmission.