Error Resilient Video Coding using Cross-Layer Optimization Approach

Abstract

A communication system can be modeled as Open Systems Interconnection (OSI)-7 layers. Generally, each layer solves its own optimization problem. For example, video coding of the application layer minimizes distortion with or without considering transmission errors for given bit rate constraints. The Transmission Control Protocol (TCP) layer solves fair resource allocation problems given link capacities, and the Internet Protocol (IP) layer minimizes the path cost (e.g. minimum hop or link price). In the Media Access Control (MAC) layer, throughput is maximized for given bit error probability, and the Physical (PHY) layer minimizes the bit error probability or maximizes the bit rate (e.g. diversity and multiplexing of Multiple Input Multiple Output (MIMO)). However, all the other layers except the application layer are not directly observed by end users. It means that the users evaluate a communication system based on quality of the application layer. Therefore, we need to maximize quality of the application layer cooperating with the other layers since one layer’s optimization can affect performance of the other layers. In this chapter, we focus on trade-offs between rate and reliability for given information bit energy per noise power spectral density Eb N0 (i.e. Signal-to-Noise Ratio (SNR)) with consideration to error resilient video coding feature. Especially, the application oriented cross-layer optimization is considered for transmission of compressed video streams. For the cross-layer optimization, the basic framework of Network Utility Maximization (NUM) (Kelly et al. (1998)) or extend framework of NUM (Chiang et al. (2007)) can be used. Especially, Lee et al. (2006) incorporate trade-offs between rate and reliability to the extend NUM framework. This framework is applied to decide the number of slices, source code rate, channel code rate, MAC frame length and channel time allocation for multiple access among