On Superposition Coding for Wireless Broadcast Channels

This thesis investigates the performance of broadcast systems in which the transmitted signal is obtained by superposition of the independent user signals. We first focus on single-input single-output systems in additive white Gaussian noise channel. We study the performance of a superposition coding based transmission scheme for a 2-user broadcast system when the power allocated to each user is varied. Using a simple scheme, we achieve near-capacity performance. When the power is split equally between the users, superposition coding leads to severe erasure at the transmitter. It is seen that this degrades the performance significantly. Next we focus on multiple-input multiple-output broadcast systems. We compare the performance of a transmission scheme based on superposition of user signals with a orthogonal transmission scheme. By using an outer turbo code and performing iterative joint detection and decoding, it is seen that linear dispersion code clearly outperforms orthogonal code. This improved performance is however obtained at the cost of increased receiver complexity. We finally study some receiver structures with reduced complexity. It is also seen that the labelling map used to map the bits to symbols has a significant impact on the performance.