Research On Cybersecurity Model And Algorithm In Dispersed Computing Environment

Dispersed computing is a new resource-centric computing paradigm,which connects all devices with computing power in the world.When computing tasks appear,it uses idle computing resources in the network to divide tasks.Dispersed computing breaks through the traditional server-client model,and provides services to users in the way of collaboration and sharing of computing nodes.Any node can be either the task publisher or the service provider.Due to its unique architecture advantages,dispersed computing can play a huge advantage in the application of the Internet of Things.While promoting the widespread implementation of dispersed computing,the cybersecurity challenges it faces must be urgently solved.First,static and single factor identity authentication is difficult to satisfy the security requirements in dynamic networks.Second,the heterogeneity,autonomy,opportunism and mobility of nodes lead to more uncertain and uncontrollable interactions among nodes,which makes it difficult to establish trust relationship and transaction rules.Third,in dispersed computing environment,nodes are widely distributed,the temporary communication links between nodes are easy to expose,and data transmission is more vulnerable to malicious interception and eavesdropping.Finally,it is difficult to balance consideration to the efficient task allocation and node location privacy protection in dispersed computing.Aiming at the main challenges of the above four aspects,this thesis conducts a systematic study on the cybersecurity model and its algorithm in dispersed computing environment,the main work and results are summarized as follows.1)This thesis establish a control model combining continuous identity authentication and security threat assessment,which comprehensively considers the two factors of security threat and control cost.Through the functional linkage of continuous identity authentication and security threat assessment,the first protection of network security is constructed.Based on the characteristics of network attacks,a security threat assessment model based on the spread mechanism of infectious diseases is established.In order to monitor the security threat index of the network in real time,a multi factor continuous identity authentication method is further proposed as an important control strategy of the assessment model.Through experimental verification,the continuous identity authentication proposed in this thesis can reduce the proportion of malicious nodes in the network from 32.5%to 1.3%in the steady state.2)This thesis establish an evolution game model of transaction strategy based on trust evaluation,which provides a trust basis and transaction rules for collaboration and sharing between nodes in a dispersed computing environment,and can induce nodes to take honest transactions as the optimal strategy to build a trusted environment.Through the theoretical analysis of the equilibrium of the model,the conditions for honest trading as an evolutionary stability strategy are obtained.In order to obtain the trust value accurately,a comprehensive trust dynamic evaluation algorithm is proposed,which is an important method to complete trust evaluation,identify malicious attacks,and eliminate malicious nodes in dispersed computing.Through the experimental verification,the interaction success rate of the comprehensive trust evaluation algorithm proposed in this paper can reach 96%,and the malicious node identification rate can reach 94.4%.3)This thesis establish a data security transmission model based on fractional order chaos synchronization is established to ensure the security of data transmission between nodes in dispersed computing environment.Based on the principle of chaotic secure communication,a data security transmission model with one to many and one to one transmission functions is established,and the corresponding algorithm is proposed.In the one-to-one data transmission model,fractional Lorenz chaotic system and fractional Rossler chaotic system are selected as the drive system at the sending end and the response system at the receiving end,respectively,to form a fractional order chaotic secure communication system with different structures and orders.Aiming at the model,a data security transmission algorithm based on chaos synchronization is proposed,and the convergence and security of the algorithm are proved.Through the calculation of the key space,the size of the key space of the communication model in this paper is obtained as 2460.4)This thesis establish a task allocation model with location privacy protection,to achieve secure,trusted,and differentiated task allocation in dispersed computing.The model comprehensively considers three optimization objectives:task completion rate,response time and communication distance.At the same time,the plane Laplace mechanism is used to provide differential privacy protection for nodes.In order to solve this model,this thesis proposed a many to many matching algorithm,and obtained a unique and stable matching.The convergence and stability of the algorithm and the optimality of the matching result with respect to the task node are proved.Through experimental demonstration,when NCPs and task nodes reach 1:2,the task completion rate of this task allocation model can reach 99.33%.