Scalable Learning-Based Heterogeneous Multi-Band Multi-User Cooperative Spectrum Sensing for Distributed IoT Systems

Abstract

The emerge of Internet of Things (IoT) brings up revolutionary changes to wireless communications. Cognitive radio (CR) can be seen as one of the prominent solutions to spectrum scarcity in IoT, where multi-band cooperative spectrum sensing (CSS) is the key. However, lack of centralized control and increase in number of devices place a room for many challenges. One of the main challenges is secondary users' (SUs') scheduling to sense a subset of channels in heterogeneous distributed CR networks (CRNs). To overcome the aforementioned challenge, in this paper, we propose a novel heterogeneous multi-band multi-user CSS (HM2CSS) scheme. The proposed scheme allows heterogeneous SUs to sense multiple channels and consists of two stages. We formulate a mathematical model to optimize leader-selection for each channel in the first stage. We then formulate another optimization problem to determine corresponding cooperative SUs to sense these channels in the second stage. After that, diffusion learning is used to decide on the availability of channels. Simulations illustrate that the proposed scheme improves detection performance and CRN throughput, is scalable in terms of detection performance, and provides fair energy consumption for CSS on all channels compared to existing multi-band CSS schemes.