Machine Learning-based Primary User Emulation Attack Detection In Cognitive Radio Networks using Pattern Described Link-Signature (PDLS)

A machine learning (ML) framework is proposed for primary user emulation attack (PUEA) detection in Cognitive Radio Networks (CRN). The ML framework is based on various classification models that exploit features extracted using the proposed Pattern Described Link-Signature method (PDLS) to distinguish between legitimate and malicious users. PDLS defines Link-signature from the channel impulse response (CIR) of the wireless link in the multipath environment. Previous works define Link-signature as the amplitude (or power) value of CIR while PDLS, on the other hand, defines it as a pattern that describes the structure of 52 sub-CIR (i.e., CIR contains 52 sub-CIR) in Orthogonal Frequency Division Multiplexing (OFDM) based transceivers. The proposed scheme is tested by developing a Software-Defined Radio (SDR) testbed to capture real wireless channel measurements. The testbed is based on IEEE 802.11a/g/p standard transceiver which comprises OFDM physical layer. Experimental results show that the proposed approach can distinguish between legitimate and malicious users effectively.