Spectrum and power optimisation in wireless multiple access networks.
Doctoral thesis, UCL (University College London).
Emerging high-density wireless networks in urban area and enterprises offer great potential to accommodate the anticipated explosion of demand for wireless data services. To make it successful, it is critical to ensure the efficient utilisation of limited radio resources while satisfying predefined quality of service. The objective of this dissertation is to investigate the spectrum and power optimisation problem for densely deployed access points (APs) and demonstrate the potential to improve network performance in terms of throughput and interference. Searching the optimal channel assignment with minimum interference is known as an NP-hard problem. The increased density of APs in contrary to the limited usable frequencies has aggravated the difficulty of the problem. We adopt heuristic based algorithms to tackle both centralised and distributed dynamic channel allocation (DCA) problem. Based on a comparison between Genetic Algorithm and Simulated Annealing, a hybrid form that combines the two algorithms achieves good trade-off between fast convergence speed and near optimality in centralised scenario. For distributed DCA, a Simulated Annealing based algorithm demonstrates its superiority in terms of good scalability and close approximation to the exact optimal solution with low algorithm complexity. The high complexity of interactions between transmit power control (TPC) and DCA renders analytical solutions to the joint optimisation problems intractable. A detailed convergence analysis revealed that optimal channel assignment can strengthen the stability condition of TPC. Three distributed algorithms are proposed to interactively perform the DCA and TPC in a real time and open ended manner, with the ability to appropriately adjust power and channel configurations according to the network dynamics. A real network with practical measurements is employed to quantify and verify the theoretical throughput gain of their integration. It shows that the integrated design leads to a substantial throughput improvement and power saving compared with conventional fixed-power random channel allocation system.
|Title:||Spectrum and power optimisation in wireless multiple access networks|
|Additional information:||Authorisation for digitisation not received|
|UCL classification:||UCL > School of BEAMS > Faculty of Engineering Science > Electronic and Electrical Engineering|
Archive Staff Only