Fast Power Control to Counteract Rayleigh Fading in Cellular Radio Systems

Sammanfattning: In urban areas, cellular radio signals are subject to Rayleigh fading, which results in rapid changes in the channel gain. For TDMA-systems without the possibility of frequency hopping, this can cause severe problems for slowly moving cellular phones.This thesis studies the possibility to use fast power control to counteract the problems introduced by Rayleigh fading in the up link. Simulations with two mobiles, with restrictions of power levels and rate of power change are made.A two mode power control algorithm basing its control channel gain predictions is investigated. An important part deals with the problem of modeling the gain of a Rayleigh fading channel using regression models, subspace models, and neural network models. Subspace models show the best performance in the noise free case but are very sensitive to noise. Combined with the power control algorithm, the choice of models seems to be of little importance when the noise level is low. The bit-error-rate in simulations is significantly reduced for low noise levels compared to a balanced strategy giving both mobiles equal Carrier-to-Interference-ratio. A comparison with the optimal solution, derived with dynamic programming under the power constraints, indicates that the algorithm is reasonably robust with respect to low noise levels. For higher noise levels, recursive autoregression models of low orders seem to be the most robust ones while subspace models and neural network models are the least robust.