The Detection Scheme Against SSDF Attacks In Industrial Cognitive Radio Network

Compared to traditional wired industrial network,industrial wireless network can be deployed flexibly in high temperature,high humidity,corrosive environments to achieve industrial monitoring in harsh environments.The characteristics of high real-time and large amount of data,the existing ISM band are scare for the demand.It can be solved by the usage of industrial cognitive radio network.In this paper,we take Bao Steel's 2050 mm hot rolling process as an example.It can be used to monitor the roughing and finishing rolling.Due to the openness of wireless network,its safety should be considered.Considering the cooperative spectrum sensing process,the second users compete for the spectrum acess.Malicious users may intend to disturb the communication of PU by launching the SSDF at the cooperative spectrum sensing process.In this paper,we study the detection scheme against SSDF attacks in industrial cognitive radio network,the main contribution as follows:An optimal reputation-based detection scheme against SSDF attacks in industrial cognitive radio network is introduced in this paper.The traditional detection scheme doesn't take the probability of attack,and the separation of malicious users and honest users is poor when the attack probability is small.We set an optimal threshold of reputation value which is designed to separate the malicious users and honest users to the greatest extent.And the threshold is closely related to the attack probability,so we apply maximum likelihood estimate method to estimate the probability of attack.And we achieve the purpose of defend against smart attackers.Simulation result reveals that our scheme performs better than existing scheme especially when the attack probability is small.Distributed network may suffer from SSDF attacks at the stage of CSS process and iteration process.We establish the attack model and analyze the effect.We introduce the outlier detection method,and the threshold updates as the iteration process going.To reduce errors caused by limited numbers of local users,we take the reputation value as the weight to participate in the consensus iteration process.The reached consensus result is good when malicious users exist.