An Intelligent Cost-Efficient Skipping Based Sensing Scheme for Cognitive Radio Transmissions

Download paper here

Based on the continuance feature of the band’s idle state observed from the result of our spectrum measurement campaign, in this paper, we propose a skipping based sensing (SBS) scheme for cognitive radio transmissions, where the sensing phase is skipped for several slots if a band is captured in the idle state, instead of sensing at each slot, so that the sensing cost could be reduced significantly. We focus on the parallel sensing scenario and consider the hardware limitation that only a portion of all available bands could be accessed. To obtain an intelligent joint decision on band selection and the number of slots to be skipped under the dynamic spectrum environment, we first develop a data-driven approach based on the partially observable Markov decision process model. Considering that the perfect knowledge on spectrum environment might be hardly obtainable in some cases, we further design a model-free reinforcement learning based method to make the decision adaptive to the spectrum dynamics. Real data is collected by using USRP-2922 to demonstrate the effectiveness of the proposed SBS scheme.

Citation: Li, X., Dong, Y., Zhang, Y., Sun, H., Zhang, R., Ding, H., An Intelligent Cost-Efficient Skipping Based Sensing Scheme for Cognitive Radio Transmissions, CN112367131A,2021-02-12