Downlink Radio Resource Allocation Schemes in OFDMA Based Cellular Systems

Resource allocation is a fundamental problem in all digital cellular systems. This thesis deals with the resource allocation problems in an orthogonal frequency-division multiple access (OFDMA) based cellular system, both at the system level and in individual cells. This thesis evaluates and compares the average capacity performance of an OFDMA based cellular system with different frequency reuse patterns and resource allocation schemes. Our work demonstrates that OFDMA cellular systems with frequency reuse factor of one can achieve higher average capacity performance than those with other frequency reuse factors for either omni-cell systems or sectorized cellular systems.;Another contribution to the system level resource allocation is two co-channel diversity schemes for users near the cell boundaries in an OFDMA based cellular system with frequency reuse factor of one. The proposed space-time co-channel diversity (STCD) and phase compensated co-channel diversity (PCCD) schemes enhance the useful signal effectively and reduce the co-channel interference simultaneously, hence improving the capacity performance for the user close to the cell boundaries significantly.;Moreover, two linear programming based and two priority-based algorithms are proposed to solve the resource allocation problem for real-time services in individual cells. The linear programming based algorithms approach the upper bound of the overall system utility satisfactorily. The priority-based resource allocation schemes perform excellently for practical OFDMA based cellular systems.;Finally, the queue-aware resource allocation problem is investigated for OFDMA based cellular systems supporting both real-time and non-real-time services. A set of practical resource allocation algorithms is proposed to solve the original non-concave integer programming problem, which converts the original problem into a minimax problem and a quota allocation problem, when necessary. Simulation results demonstrate the excellent performance of the proposed algorithms.