One important problem in spectrum sensing is to detect a noisy and unknown signal, while keeping the risk of detection error as low as possible. This problem may increase in mobile environments due to fast situation changes. In this paper, we consider a mobile cognitive radio scenario, and try to evaluate whether some knowledge about the environment and the mobility parameters of the user can help in improving the detection of changes in the spectrum occupancy. To do so, we assume that the mobility parameters can be summarized in some a priori knowledge on the average time of spectrum change and we use Bayesian changepoint detection methods. Considering that the power of the signal to be detected is usually unknown, a low-complexity algorithm is proposed that does not rely on this knowledge. It is then compared with the existing algorithms in the literature. Finally, a new metric is introduced to jointly evaluate the costs of interference and spectrum waste induced by the changepoint detection algorithms, in a time-limited communication context. Results reveal that the derived algorithm outperforms its non-Bayesian equivalent at low signal to noise ratio (SNR).
Use of Bayesian Changepoint Detection for Spectrum Sensing in Mobile Cognitive Radio
2021-04-01
3491790 byte
Conference paper
Electronic Resource
English
PLISS: labeling places using online changepoint detection
British Library Online Contents | 2012
|Cooperative Spectrum Sensing in Cognitive Radio Systems
IEEE | 2008
|