Repeated Games for Inter-operator Spectrum Sharing

TL;DR

This paper models inter-operator spectrum sharing as a noncooperative game, proposing distributed strategies to improve spectrum efficiency and QoS without revealing private network information.

Contribution

It introduces two novel game-theoretic approaches for spectrum sharing, utilizing virtual carrier pricing and mutual history, to enhance resource allocation among competing operators.

Findings

Game-theoretic approaches outperform static allocation methods.

Distributed strategies improve spectrum utilization during load asymmetry.

Proposed methods maintain fairness and efficiency in spectrum sharing.

Abstract

As wireless communication becomes an ever-more evolving and pervasive part of the existing world, system capacity and Quality of Service (QoS) provisioning are becoming more critically evident. In order to improve system capacity and QoS, it is mandatory that we pay closer attention to operational bandwidth efficiency issues. We address this issue for two operators' spectrum sharing in the same geographical area. We model and analyze interactions between the competitive operators coexisting in the same frequency band as a strategic noncooperative game, where the operators simultaneously share the spectrum dynamically as per their relative requirement. If resources are allocated in a conventional way (static orthogonal allocation), spectrum utilization becomes inefficient when there is load asymmetry between the operators and low inter-operator interference. Theoretically, operators can share resources in a cooperative manner, but pragmatically they are reluctant to reveal their network information to competitors. By using game theory, we design a distributed implementation, in which self-interested operators play strategies and contend for the spectrum resources in a noncooperative manner. We have proposed two game theoretic approaches in the thesis, one using a virtual carrier price; and the other based on a mutual history of favors. The former approach takes into account a penalty proportional to spectrum usage in its utility function, whereas in the latter, operators play strategies based on their history of interactions, i.e., how well the other behaved in the past. Finally, based on the simulations, we assess the performance of the proposed game theoretic approaches in comparison to existing conventional allocations.