Department of Electrical and Computer Engineering
Ph.D. Thesis Abstract
ADAPTIVE TRANSMISSION IN FADING ENVIRONMENT
Adaptive transmission that varies the parameters at the transmitter using channel state information is a promising technique for improving bandwidth and/or power efficiency in time-varying wireless environments. To understand the full potential of adaptive transmission, the achievable throughput of adaptive transmission systems with practical constraints needs to be determined.
In this thesis, we formulate and analyze power constrained average reliable throughput maximization with a finite number of code rates/power levels and with channel uncertainty. The focus is on communication over slow fading channels where the channel does not change over the duration of a codeword. Average reliable throughput is based on the concept of information outage, which can be related to the error performance of advanced coding designs.
For an adaptive transmission system limited by encoding/decoding only with a finite choices of code rates, it is necessary to determine the finite set of code rates that enables the system to achieve maximum throughput. If the transmitted power also varies within a finite set of values, optimum designs are closely approximated by designs found by an iterative algorithm that maximizes the throughput by alternatively varying the sets of power levels and code rates. Numerical evaluations show that systems with ten power levels and ten code rates can achieve a performance within one dB of the ergodic capacity for several fading channels.
If the transmitted power can vary continuously, throughput maximizing designs can be found by a one-parameter line search algorithm. In this case, the required number of code rates can be reduced substantially. In addition, throughput maximizing designs for an arbitrary code rate set can be obtained analytically.
When perfect channel state information is not available at the transmitter, both the transmitted power and the code rate must be adjusted according to channel estimates. If the transmitted power and the code rate vary continuously, throughput maximizing designs must simultaneously satisfy two non-linear criteria and, thus, have an ordered structure. We propose an algorithm which finds either good or optimum designs. However, for a finite number of code rates, the throughput maximization can be solved in the case of a constant outage constraint.
Numerical results show that carefully designed adaptive transmission systems with a small number of code rates (power levels) or with reasonably good channel state information can achieve throughput values close to ergodic capacity.
Ph.D. Dissertation Director: Professor Roy D. Yates
Download the Thesis: PDF