Study On Algorithms Of Resisting Malicious Users' SSDF Attacks In Cognitive Radio Networks

Cognitive radio technology has entered the eyes of experts because it can improve the utilization rate of spectrum resources effectively and resolve the spectrum crisis.In order to resist malicious users' attacks in the network,reduce the impact of malicious attacks on network performance,and ensure the controllability of spectrum resources,it is necessary to study its safety and reliability.The thesis analyzed the security threats faced by cognitive radios,focused on the research of malicious users' spectrum sensing data falsification(SSDF)attacks in the spectrum sensing technology,and proposed an improved trust-value evaluation algorithm against attacks by malicious users.The algorithm can solve the problem that malicious users' transforming-identity attacks causes repetitive damage to cognitive networks effectively,and reduce the impact of the negative impact of malicious users on network performance.Moreover,the machine learning method of support vector machine was introduced to improve the defects of the algorithm.The harmfulness of malicious attacks on cognitive networks was pointed out by analyzing the cognitive cycle,working principles and potential security risks of cognitive radio networks.Through studying the attack types and methods of malicious users and analyzing the characteristics of SSDF attacks,it was pointed out that the transformingidentity attack cannot be removed completely and the characteristics of repetitive harm to the network,and the existing algorithms cannot solve the problem of transformingidentity attack.In view of the issue that SSDF attacks affect cognitive network detection performance,a trust-value evaluation algorithm based on turn and cycle against malicious user was proposed.The algorithm removed malicious users in the network gradually through continuous loops of selecting queues of spectrum sensing randomly and turning to be reporting node.It can reduce the negative impact of transforming-identity attack on cognitive networks and restores network detection performance.The simulation results show that the detection performance is superior to the existing classical algorithms in the dynamic network and large-scale malicious user attacks,and the probability of misdetection is lower in the case of malicious users changing identity attacks.Finally,in view of the problem that the trust-value evaluation algorithm needs a long period to identify malicious users,an algorithm based on support vector machine to identify malicious users was proposed.Through machine learning methods to solve this problem,the cognitive radio network's ability to learn and reason is also exerted.It also provides a new way of thinking against malicious users.The simulation verificates the trained SVM classifier can reduce the recognition time effectively and the detection performance of SVM applied to the dynamic network is further improved.