A Novel Spectrum Sensing for Cognitive Radio Networks With Noise Uncertainty

This correspondence investigates a joint spectrum sensing scheme in cognitive radio (CR) networks with unknown and dynamic noise variance. A novel Bayesian solution is proposed to recover the dynamic noise variance and detect the occupancy of primary frequency band simultaneously. The states of primary users are detected based on particle filtering technology, and then the noise parameters are tracked by using finite-dimensional statistics for each particle based on marginalized adaptive particle filtering. Simulation results are provided to validate that the proposed method can improve the sensing performance significantly and target the dynamic noise variance accurately.