Game Theoretic Mechanisms for Resource Management in Massive Wireless IoT Systems

Abstract

As a result of rapid advancement in communication technologies, the Internet of Things (i.e., ubiquitous connectivity among a very large number of persons and physical objects) is now becoming a reality. Nonetheless, a variety of challenges remain to be addressed, one of them being the efficient resource management in IoT. On one hand, central resource allocation is infeasible for large numbers of entities, due to excessive computational cost as well as immoderate overhead required for information acquisition. On the other hand, the devices connecting to IoT are expected to act smart, making decisions and performing tasks without human intervention. These characteristics render distributed resource management an essential feature of future IoT. Traditionally, game theory is applied to effectively analyze the interactive decision making of agents with conflicting interests. Nevertheless, conventional game models are not adequate to model large-scale systems, since they suffer from many shortcomings including analytical complexity, slow convergence, and excessive overhead due to information acquisition/exchange. In this article, we explore some non-conventional game theoretic models that fit the inherent characteristics of future large-scale IoT systems. Specifically, we discuss evolutionary games, mean field games, minority games, mean field bandit games, and mean field auctions. We provide the basics of each of these game models and discuss the potential IoT-related resource management problems that can be solved by using these models. We also discuss challenges, pitfalls, and future research directions.