|Ph.D Thesis||Department of Electrical Engineering|
|Supervisor:||Distinguished Prof. Shamai )Shitz( Shlomo|
|Full Thesis text|
Traditional coded communication schemes for fading channels are based on fixed rate single level coding. The maximal average achievable rate is known as the outage capacity. When perfect channel state information (CSI) is available at the receiver end only, the outage capacity may be quite far from the ultimate performance bound, namely the ergodic capacity. In this work a broadcast transmission strategy for fading channels is studied. This strategy facilitates to adapt the reliably decoded rate to the actual channel state without having any feedback link to the transmitter. This is obtained by using a multi-layer code where each layer is associated with a channel state. Maximization of the expected throughput is achieved by optimally allocating power per layer. When the number of layers is unlimited, a continuous multi-layer upper bound can be formulated. We obtain this fundamental upper bound for channels with one degree of freedom, where transmitter has imperfect channel state information (CSI), or no CSI at all. For other multiple-input multiple output channels only achievable rates are available, due to the non-degradedness nature of the MIMO channel. In a realistic communication setting, the transmitter does not have an infinite backlog of information to send. The transmitter buffer can even be empty sometimes. We consider a simple single server queue model, concatenated with a multi-layer channel encoder, and obtain tight bounds on the expected delay. Another setting considered, is the transmission of a Gaussian source, subject to the mean squared error distortion measure. The source is assumed to be encoded in a successive refinement manner, and then transmitted over the channel using the broadcast strategy. We provide a derivation of the optimal power allocation when the fading state is a continuum. There are numerous other communication settings which may benefit from incorporating the broadcast strategy. We consider only a fraction, including hybrid ARQ protocols, and a simple setting of cooperative communications. There is still a wide range of open problems, where analytic formulation of the broadcasting upper bound is unknown.