A Research Of Push Mechanism Based On Hybrid Mobile Social Networks

In the age of information explosion,the way of getting information from the traditional active searching to receiving personalized push information.There are many third-party push platforms to complete the job.The existing push mechanism requires the connections between clients and servers.This approach will bring out many problems,for instance,aggravating network traffic overhead for servers,consuming significant amounts of data for users,and worst of all,when network environment is poor,users could not receive push messages form servers.Based on this background,this thesis designs a push system based on a hybrid mobile social network architecture.First of all,this thesis proposed a Publish/Subscribe model of the push system based on the hybrid mobile social network architecture.This thesis changed traditional Publish/Subscribe model to hierarchical structure that the servers to gateway subscriber as first layer and gateway subscriber to general subscriber as second layer.The author designed the following steps to complete the push task: publish and subscribe,interest matching,node classification,data distribution form servers or gateway to subscribers.Secondly,this thesis focuses on the push mechanism in this push system.The author obtains the encounter data about any two users by processing the users' historical location information,and then based on the social network theory,the author studies how to use the social attribute data to improve the performance of the push system.The author puts forward a definition named influence value to measure the degree of importance for a node,next,the author proposed a gateway selection strategy based on the influence value and greedy algorithm.In order to reduce the network load and improve the delivery rate,the author designs two kinds of data distribution strategies which can be adapted to different needs.Then,the author implements the algorithm through Java program,selects the appropriate parameters by experiment,and using MIT Reality Mining data set to verify the validity of the strategy.Finally,this thesis given the detailed design proposal on both sides of server and client in the push system.and then this thesis realizes some functions with the existing hybrid mobile social network platform.