Learning and Game Based Spectrum Allocation Model for Internet of Medical Things (IoMT) Platform

Abstract

The Internet of Medical Things (IoMT) paradigm provides pervasive healthcare services in-home monitoring networks. Nowadays, these services play an imperative part in the life of human beings. However, excessive requirements of health services result in insufficient spectrum resources and service delays. In this study, a novel spectrum allocation scheme is proposed for the IoMT system platform. The main challenge of our scheme is to effectively share the limited spectrum resource while dynamically handling different service requests. To achieve a mutually desirable solution for multiple IoMT devices, our proposed scheme is designed as a bi-level control algorithm using the ideas of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">multi-agent reinforcement learning (MARL)</i> and the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Balakrishnan-Gómez-Vohra (BGV)</i> solution. At the first level, each IoMT device selects its salient point according to the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MARL</i> model. At the second level, the spectrum resource is distributed through the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">BGV</i> solution, which is implemented by considering the selected salient point of each device. Through the sequential interactions of intelligent devices, our bi-level control approach can effectively guide individual IoMT devices to choose cooperation strategies while optimizing the spectrum allocation process. Finally, numerical results show the effectiveness of our proposed scheme through the comparisons with benchmark protocols. We demonstrate the performance improvement of our method in terms of the normalized device payoff, IoMT system throughput and device fairness.