Spectrum Sensing Based on kNN Algorithm for 230 MHz Power Private Networks

In this work, a spectrum sensing method based on K-nearest neighbors (kNN) algorithm is proposed. The energy and cyclic spectrum parameters of primary signal are extracted and system performance is analyzed. Results show that the detection probability of the proposed kNN method has better performance than support vector machine (SVM), energy detection (ED), and cyclostationary eigenvalue detection (CD) methods. False alarm rate of the kNN algorithm is much lower comparing with other detection methods. The proposed method reduces the complexity of cyclostationary signal detection as the kNN algorithm is easy to implement based on distance division, not on category. The provided results are useful for design of 230 MHz power private networks.