Kolkata paise restaurant game for resource allocation in the Internet of Things

The Internet of Things (IoT) is a promising networking technology that will realize many innovative smart city applications. It is anticipated that the number of connected IoT devices will greatly outweigh the available communication resources, and, thus, a key challenge is to optimize the allocation of wireless resources among the IoT devices, which are often limited in their functionality. In this paper, a distributed approach is proposed for enabling IoT devices with incomplete information and multiple objectives to effectively utilize the limited communication resources. In particular, a massive IoT consisting of IoT devices with imperfect knowledge competing over limited communication resources is formulated using a novel Kolkata paise restaurant game. For the formulated game, it is shown that the socially optimal solution coincides with the Nash equilibrium. Furthermore, a learning framework is developed to enable the IoT devices to autonomously learn their equilibrium strategies to optimize their transmission. The effectiveness of the proposed scheme in increasing the number of successful transmissions as a function of the amount of available information, device density, and transmission probability is analyzed. Simulation results show that the proposed learning framework can significantly increase the percentage of communication resources used to successfully transmit by up to threefold, compared to a baseline random allocation scheme. The results also show that the proposed learning framework with imperfect knowledge is more effective in an IoT with higher device density.