Performance Comparison of Machine Learning based Multi-Antenna Cooperative Spectrum Sensing algorithms under Multi-Path Fading Scenario

Abstract

In this paper, detection performance of various Machine learning (ML) and Deep learning (DL) algorithms based cooperative spectrum sensing (CSS) methods have been compared and analyzed. The ML algorithms are such as K-means clustering, Gaussian mixture model (GMM), support vector machine (SVM), Decision Tree (DT), and the DL architectures as artificial neural networks (ANNs) and convolutional neural networks (CNNs). To evaluate the performance of CSS methods, multi-antenna multiple secondary users (SUs) and hidden node scenarios are considered in Cognitive radio (CR) network. Such scenarios for detecting the presence of PU have not been taken into account by the system models used by the current DL-based CSS models. The fusion centre collects the SU data and computes the statistical features from sensing by adopting data fusion method. The fusion centre divides sensing data collected from all SU into two clusters and computes one-dimensional feature vector, and these features are used to train the ML classifiers. In case of DL based models, the fusion centre computes covariance matrices from the sensing data collected from each SU. These covariance matrices are fed as input to DL based CSS models. The results are showing that CNN based models outperform the ANN, and other ML based models in terms of classification accuracy and probability of detection.