Federated-Learning (FL) based Spectrum Sensing (SS) method is considered for the application in future cognitive radio communication systems due to its supreme performance in changing radio environments as compared to classic cooperative or non-cooperative SS. It also avoids transferring large training datasets with high-resolution localization data. The FL algorithm is the subject of poisoning attacks that can be random or coordinated. In this paper, we first evaluate the impact of such attacks on the FL-based SS performance. Next, we propose a zero-trust method based on continuous monitoring and classification of the sensors’ models to detect attacked models. These models are then eliminated from the global model construction in FL. Our method issemi-blind, i.e., it does not require an apriori knowledge of who are the genuine actors participating in FL. Simulation results of the system under various attacks (random or coordinated, moderate or very aggressive, deliberately increasing or decreasing the spectrum occupancy) show that our method decreases the SS probability of false alarms by 89 % and increases the SS probability of detection by 16 % in case of the most severe targeted attacks in the most critical SNR ranges.

In this paper, we present a novel approach to spectrum sensing (SS) and spectrum future occupancy prediction (SP) in a multipath, frequency-selective fading radio environment. In conventional methods, fading effects in the link between a primary user (PU) and the sensor may impede the detection of the PU transmission. Additionally, in the presence of multiple PUs and links characterized by different signal-to-noise (SNR) ratios, reliable SS and SP become even more difficult. In this paper, we look at the methods to resolve these problems. Our proposed algorithm employs machine learning (ML) methods to uncover time and frequency patterns in the occupied spectrum and occurring frequency-selective fading effects. It exploits two convolutional neural networks (CNNs). One is for Spectrum Sensing (SS) and Sensing Prediction (SP), and the other estimates fading level of a signal in a given moment. Both CNN's decisions are combined into one joint decision with improved reliability. The fading estimation provides information on the level of channel distortion that impacts the quality of SS and SP performance. The obtained results are promising and prove the usefulness of the ML-based fading-level estimation in SS and SP.