UBC Theses and Dissertations
Optimal power control for network-centric and user-centric wireless networks in interference-limited fading channels Paul, Piplu Rani
Efficient allocation of transmitter power is of paramount importance in wireless networks for both longer battery life of the mobile devices and increased utilization of the scarce wireless spectrum. Traditional power control scheme updates power whenever the fading state of the channel is changed. This approach consumes a lot of signal processing energy and may be impractical for fast-fading channels. Alternative approach is to take the statistical variation of the signal-to-interference and noise ratio of each transmitter/receiver pair into account and allocate power to optimize outage, power or utility. We address the problem of optimal power control for interferencelimited wireless networks with both Rayleigh faded desired and interference signals assuming latter approach. Unlike most of the works in the literature that use complex non-linear optimization techniques or approximate heuristic- based methods, we propose simple methods to solve the optimal power control problems. We formulate the problems from the viewpoint of both user and network. In outage-based formulations, we minimize the worst outage probability over all transmitter/receiver pairs. In utility-based formulations, which are more suitable for wireless data networks, we consider the problem of maximizing minimum utility over all transmitter/receiver pairs for network-centric scheme and individual utility for user-centric scheme. In all the schemes, we put non-negativity constraints on all the transmitted powers. We also formulate problem that minimizes the total power with specified bounds on the individual outage probability. With appropriate transformation techniques, we convert the complex constrained optimization problems into equivalent unconstrained problems, which are suitably solved using either DFP method or BFGS method. We perform extensive numerical simulations, which reveal that our proposed algorithms are very efficient to converge to the optimal solution with few iterations.
Item Citations and Data