Machine Learning based Spectrum Sensing and Distribution in a Cognitive Radio Network

The advent of 5G communication networks will seek an increasing demand in the utilization of the spectrum with the rise in the number of users. The radio spectrum is a scarce resource that must be managed and exploited effectively without sacrificing the quality of service during operation. For this purpose, the spectrum resources can be dynamically shared amongst all the users wherein the unutilized licensed spectrum of the primary users can be allocated to an unlicensed secondary user when it is underutilized by the former. This concept is called Cognitive Radio in a communication network. In this manner, the on-demand resources can be constructively taken advantage of thereby, helping the unlicensed users to be part of the spectrum at regular intervals during unoccupied times. This task is successfully achieved through spectrum sensing by the unlicensed secondary user. Hence, spectrum sensing is an important task and proper care must be taken while choosing the schemes for the same. The same concept is further extended by a real-time scenario with multiple secondary users. Here, a central hub called the Fusion Centre connects, manages, and helps in the intelligent allocation of resources amongst the secondary users. This is called Cooperative Spectrum Sensing. The use of Machine Learning will be highly beneficial. So, the addition and implementation of appropriate Machine Learning algorithms will help in intelligently predicting the spectrum allocation methods. In this work, Spectrum detection methods namely the Cyclostationary method (for High Noise environment) and Energy Detection method (for Low Noise environment) have been implemented. Machine Learning Classification techniques namely Decision Tree Classification and Random Forest Classification have been implemented for predicting the spectrum sensing method by having the received signal's Energy level (dB) and Noise level (dB) as features. A resource allocation algorithm named priority-based resource allocation has been used to allocate the free spectrum to the secondary users.