Power adaptation strategies for delay constrained channels

This thesis explores the effect of a delay constraint on optimum transmission strategies in a fading communication channel. The delay constraint is expressed in terms of the channel fading rate rather than the symbol rate. A block fading channel model is considered, which implies that the data processing occurs in (a small number of) K blocks, each block of (large) length T₀. A general cost function μ(x) is considered, along with two kinds of power constraints, the short term and the long term constraints. A causal transmit power adaptation is required to solve the problem, which is optimally computed using dynamic programming.; The general cost function is then specialized to the case of expected capacity. Expected capacity is the maximum possible ensemble-average data rate. Simulations show that optimizing the transmit power provides a substantial gain at low SNRs, but not at high SNRs. At low SNRs, a $logKm$ factor increase in capacity is shown for a $c22m$ (Nakagami) fading channel.; Outage capacity, defined as the maximum error-free data rate that can be supported at a given outage probability, is considered next. Simulations show that the long term constraint solution provides a substantial SNR gain at both low and high SNRs. The outage probability in the long term case is shown to be related to the SNR at least exponentially. Random coding bounds are derived for the outage capacity case.; A stationary version of the outage probability problem is considered, which uses an exponential window to weigh the past data rates to approximate a K block window. The optimal solution, involving linear programming, gives an SNR gain.; Space-time codes are considered as an example of the use of outage probability. In particular, an outdoor wireless multi-antenna transmission system is considered. It is shown that maximum diversity and SNR gain can be obtained by simply combining space-time codes with an appropriate ‘beamformer’.; Finally, the problem of blind symbol synchronization in OFDM is considered. It is shown that the ranks of certain autocorrelation matrices contain information that can be used to blindly synchronize the received signal, even in the presence of multipath, with an asymptotic guarantee of correctness.