Research On Defense Scheme For Byzantine Attack In Distributed Cooperative Spectrum Sensing

Collaborative spectrum sensing is an important technology in cognitive radio networks,which can greatly improve spectrum sensing accuracy in the scenarios with shadowing and fading.However,the cooperative and distributed natures of collaborative spectrum sensing make it extremely vulnerable to malicious attacks,which will make the system unstable.In this paper,the relevant defense scheme s for Byzantine attack are designed in distributed collaboration spectrum sensing,the details content as follows:Firstly,the relevant theoretical parameters about spectrum sensing are described.Some commonly used data fusion algorithms in centralized and distributed cooperative spectrum sensing are introduced.At the same time,Byzantine attack modes and corresponding security defense mechanisms in cooperative spectrum scenario are discussed.Secondly,some commonly Byzantine attack types in distributed cooperative spectrum sensing are studied.And a defense scheme in which the utility model is used in the Consensus Algorithm is proposed.In this scheme,cognitive users(CUs)are considered to be intelligent in the process of spectrum sensing,they have the ability of calculating their own utilities,and they pursue their utility maximization.Malicious users(MUs)are guided to report honestly through adjusting penalty and incentives without identifying attackers directly,and ultimately make the entire network to reach a final convergence results.Both theoretical analysis and simulation results indicate that the proposed scheme can effectively defend against various malicious attacks.Compared with the existing several defensive schemes,the proposed scheme has better security performance.Finally,due to MUs have the changeable characteristic in distributed cooperative spectrum sensing,an intelligent security mechanism combining reputation model with reinforcement learning is proposed.In this scheme,each C U is seen as an agent,they can select the cooperative users from their neighbors through reinforcement learning,reputation value is used to punish and reward cooperative users.The CUs can find out the potentially MUs through continuous learning,and make the entire network to reach a consensus results.As represented by the numerical simulation,it proves that the proposed scheme can effectively identify the malicious users,and with the deepening of the learning process,the performance of the security system will be strengthened.