Research On Detection For Primary User Emulation Attack Based On Channel Characteristic In Cognitive Radio Networks

In order to mitigate the increasingly serious problem of spectrum resource scarcity,dynamic spectrum access has been proposed to utilize the spare spectrum in the cognitive radio network(CRN).The prerequisite of this mechanism is that the secondary user(SU)cannot interfere with the primary network,which introduces new security issues for the CRN.Primary user emulation attack(PUEA)is one kind of typical attacks in the CRN,where the malicious user emulates the characteristics of a primary user(PU)with the purpose of misleading SU(s)to believe that the spectrum band has been occupied and deterring SU(s)from accessing the spare spectrum.Since the PUEA interferes with the CRN,it is necessary to investigate the countermeasures against the PUEA.In this paper,we study the detection techniques against the PUEA based on the channel characteristics.We study the PUEA detection problem with the prior information of the PU,and propose a PUEA detection method based on the multipath delay of the wireless channel.In the proposed method,the multipath delay of the fading channel is utilized as the test statistic.The prior information of the PU's characteristics and the pre-defined detection threshold are used to construct the binary hypothesis test for detecting the PUEA.Moreover,the performance of the proposed detection method in terms of the false positive probability and the false negative probability is analytically evaluated.In addition,we validate the proposed method using the experimental platform built by the universal software radio peripherals.Simulation and experimental results show that the proposed multipath-delay-based PUEA detection method achieves a high detection probability,as well as a low false positive probability.We study the PUEA detection problem without the prior information of the PU and under a variant environment,and propose a PUEA detection method based on the reinforcement learning.First,we define the network models with one PU and multiple PUs scenarios,respectively,and address the workflow of PU(s),SUs and MUs with the reward and punishment mechanism.Then,for two network scenarios,the SUs' reward is analyzed and a Q-learning-based PUEA detection method is proposed.In the proposed method,the multipath-delay is taken as the state,the detection threshold is taken as the action,and the long-term expected profit(the reward feedback to the SU)is taken as the optimization objective,the goal of the proposed method is to find the optimal threshold in different network scenarios.Simulation results show that the Q-learning-based PUEA detection method can adaptively adjust the detection threshold according to the variant environment,which results in a better detection performance,and the detection reward improvement.