Resource Allocation via Sum-Rate Maximization in the Uplink of Multi-Cell OFDMA Networks

TL;DR

This paper addresses sum-rate maximization in uplink multi-cell OFDMA networks by proposing a hierarchy of resource allocation schemes, from bounds to low-complexity practical algorithms, with performance close to optimal in simulations.

Contribution

It introduces new bounds, a centralized sub-optimal scheme, and practical low-complexity algorithms for resource allocation in multi-cell OFDMA uplink networks.

Findings

Proposed low-complexity schemes achieve near-optimal performance.

Bounds provide insights into the performance gap with the optimal solution.

Simulation results validate the effectiveness of the proposed algorithms.

Abstract

In this paper, we consider maximizing the sum-rate in the uplink of a multi-cell OFDMA network. The problem has a non-convex combinatorial structure and is known to be NP hard. Due to the inherent complexity of implementing the optimal solution, firstly, we derive an upper and lower bound to the optimal average network throughput. Moreover, we investigate the performance of a near optimal single cell resource allocation scheme in the presence of ICI which leads to another easily computable lower bound. We then develop a centralized sub-optimal scheme that is composed of a geometric programming based power control phase in conjunction with an iterative subcarrier allocation phase. Although, the scheme is computationally complex, it provides an effective benchmark for low complexity schemes even without the power control phase. Finally, we propose less complex centralized and distributed schemes that are well-suited for practical scenarios. The computational complexity of all schemes is analyzed and performance is compared through simulations. Simulation results demonstrate that the proposed low complexity schemes can achieve comparable performance to the centralized sub-optimal scheme in various scenarios. Moreover, comparisons with the upper and lower bounds provide insight on the performance gap between the proposed schemes and the optimal solution.