Power Allocation Games for MIMO Multiple Access Channels with Coordination

TL;DR

This paper uses game theory to analyze decentralized power allocation in MIMO multiple access channels, introducing a coordination mechanism to improve network efficiency and studying the existence, uniqueness, and performance of Nash equilibria.

Contribution

It presents a novel game-theoretic framework for decentralized power allocation with a coordination signal in MIMO MACs, analyzing equilibrium properties and performance gaps.

Findings

Nash equilibrium exists and is unique under certain conditions.

Coordination improves sum-rate efficiency.

Performance gap between decentralized and virtual MIMO systems is often small.

Abstract

A game theoretic approach is used to derive the optimal decentralized power allocation (PA) in fast fading multiple access channels where the transmitters and receiver are equipped with multiple antennas. The players (the mobile terminals) are free to choose their PA in order to maximize their individual transmission rates (in particular they can ignore some specified centralized policies). A simple coordination mechanism between users is introduced. The nature and influence of this mechanism is studied in detail. The coordination signal indicates to the users the order in which the receiver applies successive interference cancellation and the frequency at which this order is used. Two different games are investigated: the users can either adapt their temporal PA to their decoding rank at the receiver or optimize their spatial PA between their transmit antennas. For both games a thorough analysis of the existence, uniqueness and sum-rate efficiency of the network Nash equilibrium is conducted. Analytical and simulation results are provided to assess the gap between the decentralized network performance and its equivalent virtual multiple input multiple output system, which is shown to be zero in some cases and relatively small in general.