Resource Allocation in OFDMA Relay-Enhanced Cellular Networks

The rapid development of audio and video applications such as Skype and YouTube increases people’s demands for ubiquitous high-data-rate coverage. Orthogonal Frequency-Division Multiple Access (OFDMA) relay-enhanced cellular network, the integration of multihop relaying with OFDMA infrastructure, has become one of the most promising solutions for next-generation wireless communications. In a relay-enhanced cell, multiple Relay Stations (RSs) are deployed to assist transmissions between a Base Station (BS) and multiple Mobile Stations (MSs). However, the resource allocation becomes more complicated and crucial to gain the potential capacity and coverage improvements of relaying. Although many studies have been done on allocating resource adaptively in the traditional single-hop OFDMA networks, they can\u27t be applied to OFDMA relay-enhanced networks directly, since with the deployment of relays, resource allocation on different hops should cooperate to avoid data shortage or overflow in relays. In this dissertation, we aim to design efficient and feasible algorithms to allocate OFDMA downlink resources in a frame-by-frame basis for relay-enhanced cellular networks. To make the resource allocation problem tractable, we first consider a single cell without channel reuse, and suppose the basic unit for resource scheduling is a subchannel, each subchannel can be assigned to only one user during a scheduling period, and users’traffic is infinitely backlogged. Under these assumptions, we formulate the optimal instantaneous resource allocation problem with total power constraint to achieve the proportional fairness in the long term. Since the problem is a NP-hard combination optimization problem with non-linear constrains, it\u27s very difficult to find the optimal solution within a designated time by extensive searching over all possible solutions. We first propose a low-complex resource allocation algorithm under a constant power allocation named \u27VF w PF\u27. A void filling method is employed in \u27VF w PF\u27 to make full use of subchannels. Further more, we use continuous relaxation and a dual decomposition approach to solve the original optimization problem efficiently in its Lagrangian dual domain. A modified iterative water-filling algorithm \u27PA w PF\u27 is proposed to find the optimal path selection, power allocation and subchannel scheduling. Simulation results show the optimal power allocation can not gain much on system throughput, moreover, our optimization algorithms improve the throughput of cell edge users and achieve a tradeoff between system throughput maximization and fairness among users. However, if the basic unit for resource scheduling is a slot or users\u27 traffic is not infinitely backlogged, the resource allocation problem becomes more complicated thus it is difficult to find optimal solutions by using optimization approaches. Therefore, we propose two heuristic resource allocation schemes including a Centralized Scheduling with Void Filling (CS-VF) and a adaptive semi-distributed resource allocation scheme. Based on CS-VF, four representative single-hop scheduling algorithms including Round-Robin (RR), Max Carrier-to-Interference ratio (Max C/I), max-min fairness, and Proportional Fairness (PF), are extended to multihop scenarios to achieve different levels of fairness. Simulation results indicate that CS-VF is more adaptable to different traffic distributions and dynamic network topologies. On the other hand, the proposed semi-distributed resource allocation scheme consists of a constant power allocation, adaptive subframe partitioning (ASP), and link-based or end-to-end packet scheduling. Simulation results indicate that the ASP algorithm increases system utilization and fairness. Max C/I and PF scheduling algorithms extended using the end-to-end approach obtain higher throughput than those using the link-based approach, but at the expense of more system overhead for information exchange between BS and RSs. The resource allocation scheme using ASP and end-to-end PF scheduling achieves a tradeoff between system throughput maximization and fairness. Finally, we compare four relay-channel partition and reuse schemes in a multi-cell scenario from interference mitigation and throughput improvement points of view. Among these four schemes, 7-part partitioning (PF7) and 4-part partitioning (PF4) schemes mitigate co-channel interferences by relay-channel partitioning, while the other two schemes include partial reuse (PR) and full reuse (FR) schemes improve the throughput by relay-channel partition as well as reuse. Specially, the PR scheme achieves a tradeoff between spectral efficiency and outage. In conclusion, we formulate the optimal resource allocation problem under different as sumptions in OFDMA relay-enhanced cellular networks and give both theoretically and practically efficient polynomial-time solutions. From the theoretical point of view, we use optimization approaches including continuous relaxation and dual decomposition to find the jointly optimized power allocation, path selection and subchannel scheduling to achieve proportional fairness. From the implementation point of view, we propose two resource allocation architectures including a centralized allocation and a adaptive semi-distributed allocation, with which four representative single-hop scheduling algorithms are extended to achieve different levels of fairness in multihop scenarios. Simulation results show our optimization algorithms achieve a tradeoff between system throughput optimization and fairness among users. Simulation results further suggest that the heuristic algorithm PR+ASP+e2e-PF provides an efficient and feasible solution for multi-cell OFDMA relay-enhanced cellular networks