Indian Journal of Science and Technology

Abstract

Background/Objectives: Cognitive radio network is the evolutionary network to solve the spectrum scarcity problem. Primary User Emulation Attack is a major security issue that cause to fail the dynamic spectrum access of the cognitive radio network. Methods/Statistical Analysis: This paper proposes two level verification of the primary user signal to enhance the security of the spectrum sensing under PUEA attack. Support Vector Machine is used in the two level verification for detector classify the received data in to a location boundary and higher order statics at second level. Findings: At first level using the location information of the primary user, the primary user signal is verified for its validity. The SVM is first trained to learn the Primary user Location and with the help of known location of the Primary user it can be verified. Then the trained SVM is used to find the location boundary of the user that transmitted the signal. If the signal boundary is within the primary used location boundary it is concluded that the signal is from the primary user; else it is from the PUEA attacker .After verifying in the first level on the second level higher order statics values are calculated then using that as the feature vector the SVM classifier is used to classify the received signal as two class: one from original primary user and other from the PUEA attacker. This two level verification of the primary user signal will be more accurate method of PUEA attack detection comparing to the single level schemes. Applications/Improvements:This two level verification scheme improves the accuracy from 82% at Linear SVM kernel to 100% and also from 77% at RBF SVM kernel to 100%.
Keywords: Cognitive Radio Network, Dynamic Spectrum Access, Primary user Emulation Attack, Support Vector Machines