Secure Access Control Mechanism Of Users In The Mobile Crowd Sensing Network

Mobile Crowd Sensing(MCS)network aims to use mobile devices to collect environmental or users' activity data to support rich applications.The access of massive mobile users not only brings impetus to the application and development of MCS network,but also brings new security challenges.On the one hand,malicious users can destroy the confidentiality of MCS network through eavesdropping or remote control,etc.On the other hand,mobile users can destroy the integrity and availability of MCS network through malicious acts such as stealing or destroying data.Security issues have become one of the main obstacles restricting the development of MCS network.The paper studies how to promote the security of users' access to build a secure MCS network.The main works of this paper are as follows:1.Considering that traditional border-based security protection mechanisms cannot meet the security requirements of MCS network,this paper introduces the concept of zero-trust and designs a zero-trust based MCS network architecture.Meanwhile,the problems faced by the security threat detection technology of devices and access control mechanism of users are discussed in this architecture.2.Aiming at the problem of privacy leakage of device data caused by the leakage of model data during the distributed training process of security threat detection model,this paper designs a distributed training technology for security threat detection model based on homomorphic encryption.The additive homomorphic encryption algorithm is used to encrypt the updated model parameters of devices on the terminal side,which protects the transmission and aggregation security of model data.Additionally,aiming at the problem of low detecting accuracy of global model caused by the low-quality data from devices,a box-plot method is utilized to improve the quality of aggregated model by identifying and filtering outliers in the data to be aggregated.Finally,safety analysis and simulation experiments verify that the proposed scheme not only protects the privacy security of devices and model data,but also reduces the impact of low-quality data on the detection accuracy of model.3.Aiming at the problem that existing access control mechanisms,due to the single authorization mode,the dynamic authorization of users cannot be achieved in the proposed architecture,a multi-layer hybrid access control model based on roles and attributes is designed.The proposed model adds a dynamic role-based permission distribution layer and a role-centric permission filtering layer into the role-based access control model to achieve fine-grained and dynamic authorization.Additionally,considering the impact of changes in the access behavior of users on the security of resource access,the permission filtering layer introduces the concept of trust,designs a permission filtering rule based on trust constraints to adjust the users' permission according to changes in users' access behavior,improving the dynamic of access control.Finally,performance analysis and simulation experiments verify that the effectiveness of the proposed access control mechanism in terms of control accuracy and average response time.