Optimum resource allocation for heterogeneous wireless OFDM networks

The popularity of orthogonal frequency division multiplexing (OFDM) has been evident by the present and the next-generation broadband wireless standards. In OFDM systems, two principle transmission resources are provided, i.e., subcarriers and transmission power. The primal challenge for designing wireless networks is to use the transmission resources as efficiently as possible while satisfying requirements of data transmission. Future cellular networks are expected to provide a large variety of services with diverse quality-of-service requirements. Thus, this thesis investigates resource allocation for unicasting by a single or multiple Base Stations (BSs), where a mixture of real time and non-real time applications is considered, so called heterogeneous resource allocation. First, with duality theory a dual optimum of resource allocation is determined for unicasting by a single BS. Simulations demonstrate that the duality gap between the dual and primal optima decreases approximately exponentially in the number of subcarriers. Thus, the dual optimum is qualified to be a reference for evaluating heuristic solutions. To reduce the computational complexity, heuristic methods are then proposed. It is verified by simulations that the obtained suboptimal solutions are close to the dual optimum. Furthermore, two different scenarios are considered for unicasting by multiple BSs. If signals received from different BSs are synchronized, each user may be served by more than one BS. Otherwise, each user can be covered by only one BS. The dual optimal and heuristic solutions for single-cell heterogeneous resource allocation are extended to these multicell scenarios. Simulations show that the derived suboptimal solutions approach the dual optima. Moreover, a novel strategy of resource allocation is proposed, which differs from the water-filling strategy. We suggest that a rate is symmetrically allocated to subcarriers assigned to each user. In doing so, the signaling overhead is reduced and a better balance of energy consumption is achieved for signaling and data transmission. The asymptotic limits are deduced for the resulting instantaneous per-symbol performance loss. By using the simplicity of the proposed strategy, heuristic solutions are designed for heterogeneous resource allocation. With signaling overhead additionally considered, the proposed strategy achieves better energy efficiency than the water-filling solution, when the resource allocation scheme is frequently updated. Finally, we deal with two important practical issues on resource allocation. On one hand, channels are measured at receivers and then imperfect channel knowledge is fed back to the transmitter. An upper bound is deduced for resource allocation in the presence of imperfect channel knowledge. On the other hand, the transmission rate must be discrete because of a limited number of available coding and mapping schemes. A non-iterative method is proposed for optimally quantizing the output rates of a water-filling solution