Distributed Resource Allocation in Multi-Cell OFDMA Networks
【摘要】：Using the intrinsic characteristic of channel orthogonality, we emphasize on Orthogonal Frequency Division Multiple Access (OFDMA), which is the proposed modulation and multiple access method for 3G LTE wireless communication systems. Despite the fact that OFDM combats intra-cell interference inside a single cell, inter-cell interference due to frequency reuse exists and plays an important role in the multi-cell environment. In this thesis we study resource allocation methods, distributed per base station (BS) in multi-cellular OFDMA networks. The objective is to provide the Quality of Service (QoS) requested by each user, whatever its location in the cell. First, it investigates causal network coordination in distributed networks. Two BSs from a virtual Multiple-Input Multiple-Output (MIMO) array for the users located at the border of cells. These users thus benefit from a diversity gain and from inter-cell interference mitigation.
The efficiency of the associated resource allocation method depends on the fairness of the power control objective. Thus, network coordination is used for Rate Constrained (RC) users, but not for Best Effort (BE) users, in a proposed algorithm that jointly managers both QoS objectives. This work in this thesis considers the more general perspective of full distributed networks. For RC users, a resource allocation process with iterative interference-based power allocation is determined to solve the Margin Adaptive problem. It includes a distributed constraint that guarantees power control convergence. The proposed method is extended to RC users in MIMO, both when full Channel State Information is available at transmission and when only the statistical properties of the channel are available at transmission. Finally, for BE users, the objective is to maximize the weighted sum throughput, when the weight of each user is proportional to its queue length. A subcarrier allocation method, deduced from a network-wide interference graph, and a distributed power method are proposed for that optimization problem.