Abstract
Lower data rates for users with low channel quality, who may either be located further from the base station, or suffer high levels of shadowing, multi-path fading or inter-cell interference, is one of the main challenges in OFDMA based, multi-cell, wireless environments. The new flat architecture design for these networks provides new challenges and opportunities for coordination among the base stations. In these networks, the radio resource management functionalities are terminated in the base stations, which are directly interconnected through high speed, high capacity backhaul links. This thesis focuses on coordinated resource allocation, which takes advantage of the introduced backhaul links in the new architecture. The contributions of this thesis are summarised in the following. First, a novel scheme for coordination in resource allocation for a cluster of neighbouring base stations is proposed. By applying a simplified power allocation strategy to the original optimisation problem, and based on the information shared among the neighbouring cells, the scheduling decisions are made locally and independently by each cell. The proposed scheme is implemented in a distributed manner, and allows a user with low channel quality to be served temporarily by an adjacent base station, which has a better channel quality to that user for that specific time instance. Information is exchanged only once during each scheduling epoch, which results in reduced overhead on the backhaul links. Simulation based performance analysis demonstrates an average of 30% improvement in spectral efficiency compared to scenarios with no coordination. Next, a novel distributed, coordinated inter-cell frequency planning scheme is proposed. While the user differentiation is performed in the mobile station, the frequency planning is performed in a predefined interval and in a distributed manner by the base stations. In this scheme the available resources are allocated to different virtual schedulers, which in turn will perform scheduling for the users allocated to them based on the relative received signal strength. This simplifies the original, complex multi-cell resource allocation problem, into a single-cell resource allocation problem, which is solved by each virtual scheduler. Simulation based analysis shows that this scheme outperforms the static schemes in the literature by an average of 45%, and has a better performance in terms of fairness. Finally, base station coordination in a relay-enhanced, multi-cell environment is investigated. The problem is analytically solved using variable relaxation and Lagrangian dual method. A suboptimal, yet efficient collaborative resource allocation scheme is proposed, in order to provide low complexity solutions for practical implementations. The key concept behind the proposed scheme is that a dynamic, collaborative frequency planning can take place in a longer time scale while the resource allocation is performed in a shorter time scale, allowing the system to fully reap the benefits of multi user diversity and temporary channel conditions. It is demonstrated that the proposed heuristic scheme outperforms the static schemes in terms of total system throughput by ten percent, and archives close performance to that of the optimal solution and maintains a good level of fairness among the users.