Research And Application Of Dynamic Community Division Based On Mobile Social Network

Delay Tolerant Networks(DTNs),also known as Tolerant Networks,refer to a special type of network in which end-to-end paths are often difficult to establish,and message propagation in the network has a large delay Is a new type of network system with intermittent links between nodes,and mobile social networks(MSNets)is an important application of tolerant networks.Community division has also become an important concern in the delay tolerance network.Also,dynamic community detection is applied to data forwarding in delay-tolerant networks,it has received increasing attention.However,the existing community detection methods are basically aimed at static communities.However,the existing delay-tolerant network data forwarding algorithms mostly pay attention to the static attributes of the nodes when measuring their propagation capabilities,and ignore the dynamic changes of the attributes over time.Based on the above viewpoints,this article mainly elaborates on two aspects of contribution.One is a data forwarding algorithm(NC-based)based on the prediction of the change rate of node neighbors.The second is dynamic community planning based on Jaccard similarity,and a new routing and forwarding algorithm is proposed.For the first contribution,this paper studies the relationship between data forwarding and node neighbor change rate in DTNs.First,the neighbor change rate of the nodes between adjacent time periods is defined,and three prediction methods for the node's future neighbor change rate are proposed.Then,a data forwarding algorithm(NC-based)based on the prediction of the change rate of the neighbors of the nodes is proposed.The main idea is to measure the ability of the nodes to forward messages based on the changes of the neighbors of the nodes in adjacent time periods.Finally,simulation results show that compared with the existing data forwarding algorithms,the NC-based algorithm can significantly reduce the network overhead while ensuring a successful transmission rate.At the same time,the paper concludes that when the size of the time window is the average time of node contact,the overall performance of the NC-based algorithm reaches the optimal.For the second contribution,this paper proposes a location-based Jacobian similarity method for dynamic community detection(DCPLJ),which makes full use of adjacent blocks of l characters at corresponding positions in two adjacent slots Changes to detect dynamic communities.In this method,we divide a node's neighbor sequence vector into several overlapping blocks,called l-character neighbor blocks,as a basic unit to study its change over time.We explored location-based Jaccard similarity,weights,and the contribution of adjacent blocks of l characters to describe the extent to which it changes in corresponding positions in two adjacent slots.Based on this exploration,we define the location-based Jaccard similarity of the neighbor sequence vectors of the nodes,and obtain the location-based Jaccard similarity of the network topology in two adjacent time slots.We can then determine if the network topology needs to be re-divided in the next time slot.In addition,we analyzed several indicators from the perspective of the location-based Jaccard similarity of the block,including the setting of the time slot size of the divided community and the sufficiency condition of the community division.Third,we also propose a new data forwarding strategy for the delay-tolerant network,called a dynamic community-based data forwarding algorithm.The idea is to send data packets only to nodes in the same community as the target node.Finally,the evaluation results show that our strategy achieves higher packet transmission rates and smaller transmission delays in the three real traces,especially when compared to traditional community-based strategies with significantly lower overhead.