Research On Malicious Attack And Intelligent Defense Scheme In Distributed Cooperative Spectrum Sensing

Cognitive radio(CR)uses the dynamic spectrum access mode to realize the two utilization of the free spectrum of the Secondary Users,which greatly improve the spectrum utilization and solves the problem of the shortage of spectrum resources for a long time.Among them,Cooperative Spectrum Sensing(CSS)can effectively overcome the influence of multipath effect,shadow fading and transmission loss,and effectively improve the spectrum utilization.However,when the cognitive users have malicious users,they transmit false sense values to the neighbors,and mislead the neighbors to make the wrong decision,which greatly destroys the performance of the entire cognitive network,so the detection and defense of malicious users become particularly important.This article focuses on the Byzantine attack(Spectrum Sensing Data Falsification,SSDF)to propose relevant defense schemes,which are as follows:First,we introduce the relevant theoretical knowledge of cooperative spectrum sensing in centralized and distributed scenarios,such as the fusion algorithm,and the fusion criteria.In addition,the Byzantine attack in the cognitive wireless network and the corresponding defense measures are introduced in detail.Secondly,aiming at different types of Byzantine attacks,we focus on the scenario of collaborative attacks among malicious users,and propose a distributed intelligent intrusion prevention scheme based on reputation and consensus.In this scheme,reward and punishment mechanism is used to reward or punish users' reputation value in every iteration process,and the reputation values of cognitive users will be combined with the fusion factor in consensusalgorithm to achieve the consensus process.In the end,the proportion of malicious users in the process of fusion is becoming smaller and smaller,and the proportion of honest users is getting greater and greater,malicious users eventually give up malicious attacks,and began to send the correct sensing value,which achieve the consensus of full network.Through simulation analysis,the proposed scheme can effectively resist a malicious user(cooperation / non-cooperation)attacks,which greatly improve the robustness and stability of the cognitive network.Finally,a defense scheme of intelligent reinforcement learning based intrusion is proposed in this paper,which combine the reinforcement model with the reputation model,and select the optimal neighbor users to cooperative,and then get a corresponding instantaneous reward and cumulative reward,finally update the corresponding Q corresponding.At the same time,we updatethe reputation value.When the neighbors whose reputation value is less than a threshold value is regarded as potential malicious users,the intelligent malicious users finally give up the malicious attacks and send the correct sensing values,and finally achieve the consensus of the whole network.Through the simulation analysis,the scheme can effectively resist the scenes of multiple malicious user attacks and greatly improve the robustness of the whole cognitive network.