Cooperative spectrum sensing in Cognitive Radio: An Archetypal Clustering approach

Cognitive Radio (CR) is an intelligent wireless communication system capable of sensing the environment and making decisions on how to use the available radio resource without creating any harmful interference to licensed users (Primary Users). The intelligent system module of CR provides the ability to gain the knowledge of available spectrum opportunities and reconfigure Radio Frequency (RF) operating parameters. Spectrum Sensing is a key mechanism in CR for acquiring spectrum awareness. The performance of spectrum sensing algorithm degrades due to channel impairments, such as, multipath fading, correlated shadowing and receiver uncertainty issues. To overcome this limitation, Cooperative Spectrum Sensing (CSS) was introduced to take the advantage of spatial diversity of wireless receivers. In recent years, cooperative sensing based on Machine learning has been used to improve the efficiency of learning in CR. Cooperative learning can help a CR to learn the surrounding environment and improves sensing accuracy. In this paper, we propose unsupervised Archetypal Clustering approach in which the local energy vectors are decomposed into a collection of extreme points which are called Archetypes. Archetypal Clustering has similar flavor to the well-known K-means clustering, while the later provides an average view on the data, the former provides an extremal view. The performance of proposed algorithm is quantified in terms of target Probability of detection and false-alarm to meet the requirement of IEEE 802.22 Wireless Regional Area Network (WRAN) standard. Further, we compare our results with k-means clustering and Support Vector Machine (SVM).