K-Nearest Neighbour (KNN) Algorithm based Cooperative Spectrum Sensing in Cognitive Radio Networks

The ability to provide dynamic spectrum access through cognitive radio technology could help address the shortage of radio spectrum. In order to allocate the free band to the secondary users, the identification of the primary user's frequency band must be performed. In order to perform the task efficiently, it is necessary that the K-Nearest Neighbor algorithm is used to classify the secondary users. This paper presents a machine learning algorithm that is designed to perform the spectrum sensing task. The performance of the KNN algorithm is analyzed in terms of its Accuracy, Precision, Sensitivity, Specificity, F1_score, Confusion Matrix, k factor by varying the percentage of test users. The K-Nearest Neighbor algorithm is designed to perform the task efficiently and effectively by selecting the optimal k value from the obtained results. It has a simple working and computation design that allows it to be modified in various ways to improve its accuracy and applicability.