The application and comparison between machine learning algorithms in cooperative spectrum sensing
Bin Hu,Yuxiang Liu,Mingxi Zhai,Aoxiang Wang
DOI: https://doi.org/10.1007/s11042-023-17692-w
IF: 2.577
2024-02-29
Multimedia Tools and Applications
Abstract:The world is progressing at a rapid rate, and with continuous advancements in technology, the need for cooperative spectrum sensing has evolved. Cooperative spectrum sensing is an approach used to enhance detection performance, wherein secondary users collaborate with each other to sense the spectrum and identify spectrum holes. As technology improves over time, the spectrum becomes increasingly allocated to primary users. The role of cooperative spectrum sensing in cognitive radio fulfills the essential requirement of protecting primary users from harmful interference. The progressive evolution of technology has led to a reduction in available spectrum, prompting the emergence of the concepts of cognitive radio and cooperative spectrum sensing. To begin with, this paper introduces the fundamental application of cooperative spectrum sensing algorithms, encompassing primary and secondary users' models and an energy model. Subsequently, three machine learning algorithms, namely K-means, support vector machine, and logistic regression, are explained, and their schematics are presented in detail. The proposed model uses supervised and unsupervised learning techniques to develop a cooperative spectrum sensing framework. It compares the performance of three machine learning algorithms, K-means clustering, logistics regression and support vector machine. The comparison is based on accuracy, recall and precision metrics, and the results show that the K-means clustering algorithm has better performance than the other two algorithms. The findings highlight the superiority of the K-means clustering algorithm over logistic regression and support vector machine.
computer science, information systems, theory & methods,engineering, electrical & electronic, software engineering