Analysis of Cognitive Radio Scenes Based on Non-cooperative Game Theoretical Modelling

TL;DR

This paper applies non-cooperative game theory to analyze spectrum access in cognitive radio environments, exploring equilibrium concepts and modeling scenarios to optimize user payoffs and interactions.

Contribution

It introduces a game-theoretic framework for cognitive radio spectrum access, incorporating multiple equilibrium concepts and realistic user behavior models.

Findings

Nash equilibrium limits channels per user without payoff loss

Pareto equilibrium allows broader, fairer payoff distributions

Stackelberg model shows optimal payoffs when incumbents are Nash-oriented and entrants are Pareto-driven

Abstract

A noncooperative game theoretical approach for analysing opportunistic spectrum access (OSA) in cognitive radio (CR) environments is proposed. New concepts from game theory are applied to spectrum access analysis in order to extract rules of behaviour for an emerging environment. In order to assess OSA scenarios of CRs, two oligopoly game models are reformulated in terms of resource access: Cournot and Stackelberg games. Five CR scenes are analysed: simultaneous access of unlicensed users (commons regime) with symmetric and asymmetric costs, with and without bandwidth constraints and sequential access (licensed against unlicensed). Several equilibrium concepts are studied as game solutions: Nash, Pareto and the joint NashPareto equilibrium. The latter captures a game situation where players are non-homogeneous users, exhibiting different types of rationality, Nash and Pareto. This enables a more realistic modelling of interactions on a CR scene. An evolutionary game equilibrium detection method is used. The Nash equilibrium indicates the maximum number of channels a CR may access without decreasing its payoff. The Pareto equilibrium describes a larger range of payoffs, capturing unbalanced as well as equitable solutions. The analysis of the Stackelberg modelling shows that payoffs are maximised for all users if the incumbents are Nash oriented and the new entrants are Pareto driven.