Recent work has suggested that adaptive signaling has the theoretical promise of greatly increasing the bandwidth efficiency of digital communication systems operating over wireless channels. However, the utility of adaptive signaling has been questioned due to the variation of the wireless channel over time, which results in a different channel at the time of data transmission than at the time of channel estimation. Recent results by the PI have shown that these issues cannot be disregarded; the channel variation greatly alters the nature of the problem and thus the design of robust adaptive signaling schemes for wireless channels presents significantly different challenges than the analogous design problem for wireline channels. Many of these challenges are driven by the robustness necessary in these algorithms due to uncertainties at the transmitter about the statistics of the channel fading process. In this project, a comprehensive study of the design of robust adaptive signaling schemes for time-varying channels is undertaken. The main portions of this study include: (1) The design and characterization of robust algorithms that employ a given adaptive signaling paradigm; (2) The development of adaptive coding paradigms that are well suited to time-varying channels; and (3) The development of adaptive hybrid-ARQ schemes, which have the added flexibility of partially guaranteeing robustness through retransmission. The coupling of this work with the development of the necessary statistical models, in particular the development of appropriate classes of autocorrelation functions for the channel fading processes, will lead to families of adaptive coding strategies that can be employed in systems operating over time-varying channels with various rates of channel variation. These results will illustrate both the applications where adaptive coding is useful and the manner in which it should be employed. The methods that will be used to obtain these results are drawn from the areas of pr obability and random processes, communication system theory, and statistical signal processing. In particular, the results rely heavily on results from random process theory, the design of coded modulation schemes, and extensions of results from minimax-robust time series prediction.
