Cooperative Spectrum Sensing algorithm based on Federated Learning for Broadcasting Services

With the further development of big data, paying attention to data privacy and security has become a worldwide issue, and every data leakage will cause great concern to the media and the public. In order to solve the problems of data privacy of individual users, data islands and ensure the quality of spectrum sensing (SS) model during migration, a cooperative SS algorithm based on federated learning (FL) is proposed in this paper, which is effective in solving the problem of SS and effectively improves the utilization efficiency of spectrum. Specifically, the users adopt the local data sets to train the convolutional neural network (CNN). Then update the trained model parameters to the fusion center that performs global aggregation. Experimental results show that the cooperative spectrum sensing (SS) based on federated learning (FL) can effectively improve the sensing performance at low signal-to-noise ratio (SNR).