Research On The Influence Maximization Algorithm Of Device-to-device Mobile Social Networks

Social networks have changed the way people communicate with each other and become an important medium for information diffusion.With the popularity of smart mobile devices,more and more people are beginning to use smart devices to access social networks,exchange information anytime and anywhere,and mobile social networks are beginning to emerge.However,mobile social networks bring great convenience to people,but also lead to the explosive growth of information,which brings huge challenges to the underlying communication network.Research shows that Device-to-Device(D2D)technology can effectively offload mobile communication network load.The study of information transmission in D2 D environment is of great significance for using D2 D technology to cope with the challenges faced by mobile communication networks.Maximization of influence is one of the important contents of information dissemination analysis.It tries to achieve the purpose of propagating information in the maximum range by finding the node with the largest influence.This is closely related to the D2 D technology to offload the mobile communication network.Based on those,this paper focuses on the issue of maximizing influence in D2 D mobile social networks.The research contents include:(1)According to the characteristics of D2 D environment,factors such as activity and closeness related to user activities are introduced into the influence computing model.In addition,invalid node elimination mechanism is added in the process of influence propagation to improve the propagation performance.(2)For the time cost and propagation efficiency caused by iterative calculation in the influence maximization,this paper proposes a constraint influence calculation model.In a limited range,the greedy strategy is applied to rapidly calculate the diffusion influence.Then the candidate seeds are selected and the overlapping cost detection is performed,the seed nodes are selected according to the detection result.(3)Performance issues caused by a large number of iterations in the linear threshold model have been improved.A prediction mechanism is added to the model,and the influence propagation result is activated one step ahead according to the neighbor relationship,which accelerates the convergence speed of the algorithm.