Automatic spectrum sensing techniques using deep convolutional neural network in cognitive radio network

Cognitive Radio (CR) network is established for spectrum utilization. This technology allows the unlicensed users to share the spectrum with licensed users. In order to perform such process, spectrum need to periodically scanned in order to find the voids in the white (licensed) spectrum. Automatic spectrum sensing approaches are proposed in this paper. Deep learning classifier namely convolutional neural network (CNN) and machine learning approaches such as Random Forest (RF), Support Vector Machine (SVM), Principle Component Analysis (PCA), K-nearest Neighbour (KNN) and Bagging algorithm. CNN based spectrum sensing is outperformed, a 86.4 % spectrum sensing accuracy is achieved using this technique.