Energy-Efficient Spectrum Sensing for IoT Devices

Device-to-device communications have been considered as an indispensable enabler, which reduces the traffic burden associated with fifth-generation (5G) mobile networks. To improve the radio spectrum utilization under such a communications scheme, cognitive spectrum sensing can be used to identify temporarily available spectrum chunks for direct interconnections among user devices. Although various sensing techniques have been proposed during the last decade, improving the sensing efficiency (SE), such as energy reduction and positive sensing ratio, remains an open challenge. The problem becomes more pronounced in 5G networks, wherein battery-constrained Internet-of-Things devices (IoTDs) are densely interconnected. In this article, we optimize the SE by leveraging a lightweight yet effective adaptive medium learning method with a probabilistic decay feature. Specifically, the wireless channels that are likely available for IoTDs are sorted and sensed in the descending order of their availability likelihood/probabilities, which indicate the estimated percentage of the availability of the sensed channels. These probabilities learn from the preceding sensing results, and they decay with time. Numerical results show that the proposed sensing approach achieves significant SE improvement compared to state-of-the-art algorithms.