The Research On Distributed Cooperative Spectrum Sensing Technology Against SSDF Attack

Cooperative spectrum sensing(Cooperative Spectrum Sensing,CSS)is always drawing much attention as one of the key technologies in cognitive radio,where the reduced sensing performance introduced by multipath effects and shadow fading can be effectively alleviated and the spectrum utilization can be improved by multi-user cooperative spectrum sensing.However,malicious users in CSS may mislead the fusion center or neighbor nodes to make wrong judgments by sending fake messages,which will greatly reduce the spectrum utilization.Based on the Byzantine attack(Spectrum Sensing Data Falsification,SSDF)and its defense scheme,we propose two solutions to defend against SSDF attacks for distributed scenario.In this paper,we first introduce the theoretical knowledge of CSS,and then we introduce our proposed intelligent solutions,as follows,Firstly,we introduce the related theoretical knowledge of CSS in centralized and distributed scenarios respectively.Specifically,in centralized scenarios,we introduce five common fusion rules.In distributed scenarios,we introduced common fusion algorithms.Furthermore,the common Byzantine attack methods and their existing defense methods are introduced.Then,in light of the widespread average consensus algorithm in distributed scenarios,we propose an intelligent SSDF attack prevention scheme based on the maximal likelihood estimation.In distributed scenarios,each cognitive user can only receive the perception values from neighbor nodes,so that the number of samples is small enough compared with that in the centralized scenario.As a result,the ability of cognitive users to identify fake perception values forged by malicious users is lower.Therefore,we propose a malicious user identification scheme based on the maximal likelihood estimation by integrating the location information of neighbor nodes,and performing the maximal likelihood estimation on the true perceived value,and further to treat neighbor nodes that deviate greatly from the estimated value as potential malicious users which are excluded in the nodes of consensus fusion.Simulation result analysis that our proposed scheme can effectively identify malicious users with a small number of received samples,so that honest nodes can converge to the correct perception value.Finally,based on the reputation and non-consensus messaging algorithm,we proposed an intelligent anti-SSDF attack algorithm,with the aim to avoid malicious nodes tamper with the content of the message when the message passing algorithm obtains the perception value of the entire network by forwarding the messages from neighbor nodes.Different from the conventional reputation value algorithm that requires the network to cooperatively perceive historical information,our proposed anti-SSDF attack algorithm uses the historical propagation path of each message as the basis for calculating the reputation of each cognitive user.Once a node receives different messages from the same cognitive user,there must be malicious users on its propagation path,where we reward the nodes that appear more frequently on different paths with reputation values and punish the nodes that appear less frequently with the reputation values.Finally,based on the calculated reputation value,we calculate the true value of the tampered message.Simulation analysis shows that the proposed algorithm has a high reduction rate for messages while identifying malicious users in the network.