Adaptive rate control for time-varying communication channels with a feedback link

• Main Author: Cavers, James Kennedy
• Language: English
• Published: University of British Columbia 2011
• Subjects: Adaptive control systems
• Online Access: http://hdl.handle.net/2429/34769

Many communication channels have a power-to-noise ratio (PNR) which is not constant in time, producing a time-varying error probability. If a feedback channel is available, the receiver can request changes in certain transmitter parameters in response to the changing channel conditions. In this thesis a scheme for adaptively changing the data rate by varying the duration of the transmitted pulses in order to compensate for this fluctuating PNR is described and analyzed. Implicit equations for the optimum rate request as a function of past and current instantaneous PNR have been derived for an arbitrary probability density function of the PNR. The effects of a bandwidth constraint, of time delay in the feedback link, and of time and amplitude discrete rate requests have been included in the analysis. Application of adaptive rate control to the Rayleigh fading channel can produce an enormous reduction in required transmitter power over a fixed rate non-diversity system, up to 50 db for typical values of error probability. For the same values of bandwidth, data rate, and error probability, and for typical values of feedback delay, the variable rate system can still effect a power reduction in the range 15-18 db, or a factor of 30-60, over the best alternative scheme, known as maximal-ratio predetection combined frequency diversity. A method is given which allows tradeoffs between power, bandwidth and data rate for two-way communication over Rayleigh fading channels to be examined graphically. Adaptive rate control on multi-user channels produces a smaller improvement. For the range of parameters considered likely, there is a maximum of about 1.7 db improvement over a fixed rate system. Although the magnitude of the improvement introduced by adaptive rate control is strongly dependent on the probability density function of the PNR, it has been shown that for at least one commonly occurring class of time-varying channels the savings are well worth the cost of implementation. === Applied Science, Faculty of === Electrical and Computer Engineering, Department of === Graduate