Distributed power allocation for D2D communications underlaying/overlaying OFDMA cellular networks

TL;DR

This paper introduces distributed power allocation algorithms for D2D communications in OFDMA networks, leveraging potential game theory to improve throughput and power efficiency in dedicated and reuse modes.

Contribution

It presents novel distributed iterative algorithms based on potential game modeling for D2D power allocation in OFDMA networks, applicable to multiple operating modes.

Findings

Algorithms converge to local maxima of the sum rate.

Performance close to optimal solutions in various scenarios.

Outperforms existing schemes in the literature.

Abstract

The implementation of device-to-device (D2D) underlaying or overlaying pre-existing cellular networks has received much attention due to the potential of enhancing the total cell throughput, reducing power consumption and increasing the instantaneous data rate. In this paper we propose a distributed power allocation scheme for D2D OFDMA communications and, in particular, we consider the two operating modes amenable to a distributed implementation: dedicated and reuse modes. The proposed schemes address the problem of maximizing the users' sum rate subject to power constraints, which is known to be nonconvex and, as such, extremely difficult to be solved exactly. We propose here a fresh approach to this well-known problem, capitalizing on the fact that the power allocation problem can be modeled as a potential game. Exploiting the potential games property of converging under better response dynamics, we propose two fully distributed iterative algorithms, one for each operation mode considered, where each user updates sequentially and autonomously its power allocation. Numerical results, computed for several different user scenarios, show that the proposed methods, which converge to one of the local maxima of the objective function, exhibit performance close to the maximum achievable optimum and outperform other schemes presented in the literature.