Bandwidth and rate allocation tradeoffs of source : channel coding, packetization and modulation in unequally protected multimedia communication systems

A conventional approach to the design of wireless multimedia communications is the layered approach, in which the network layers function independent of each other. This kind of layered approach is inspired partly by Shannon's separation theorem in which the optimization of each block is equivalent to optimization of the overall source-channel coding operation. However, the separation theorem is valid only in a point-to-point communication scenario in the case of asymptotically long block lengths of data and assumes huge amounts of processing power and delay. Therefore, current practical communication systems strive to jointly design building blocks of a multimedia system for better performance. The focus of this dissertation is therefore to present various joint designs for different channel models and systems, although limited by physical constraints such as bandwidth, power and complexity. First, a robust coded scheme for progressive multimedia transmission is considered for an additive white Gaussian noise channel, a Rayleigh fading channel, and a frequency selective channel using in combination different unequal protection methods. We investigate the judicious use of the limited bandwidth through the combination and optimization of a progressive source coder, a rate compatible punctured convolutional code and a hierarchical modulation. Next, we investigate a novel packet formatting scheme for progressive sources using interleavers and various channel codes. The source coder is combined with a concatenated block coding mechanism to produce a robust transmission system for embedded bit streams. The objective is to create embedded codewords such that, for a particular information block, the necessary protection is obtained via multiple channel codings, contrary to the conventional methods which use a single code rate per information block. We show that near capacity performance can be achieved using the proposed scheme in conjunction with low density parity check codes in a binary symmetric channel scenario. We initially focus on coding strategies for multimedia where the channel state information is missing. A generalized Digital Fountain (DF) code is proposed to provide efficient universal forward error correction solution for lossy packet networks with increased unequal error protection and unequal recovery time properties. We propose a progressive source transmission system using this generalized code design. We apply the generalized DF code to a progressive source and show that it has better unequal protection and recovery time properties than other published result