6533b836fe1ef96bd12a1c45

RESEARCH PRODUCT

Proactive Handoff of Secondary User in Cognitive Radio Network Using Machine Learning Techniques

subject

description

Spectrum management always appears as an essential part of modern communication systems. Handoff is initiated when the signal strength of a current user deteriorates below a certain threshold. In cognitive radio network, the perception of handoff is different due to the presence of two categories of users: certified/primary user and uncertified/secondary user. The reason for the spectrum handoff arises when the primary user (PU) returns to one of its band used by the secondary user. The spectrum handoff is of two types: reactive handoff and proactive handoff. There are certain limitations in reactive handoff, such as it suffers from prolonged handoff latency and interference. In the proactive handoff, the operation of handoff is planned and implemented by predicting the emergence of primary user based on the historical data usage. Therefore, proactive handoff boosts the performance of a cognitive radio network. In this work, a spectrum prediction technique is proposed for ensuring the spectrum mobility using machine learning. Machine learning techniques such as decision tree, random forest, stochastic gradient classifier, logistic regression, multilayer perceptron, and support vector machine are researched and implemented. The performance of different techniques is compared, and the accuracy of prediction is measured.