基于社交圈划分与相遇时间预测的机会网络路由算法

Routing Algorithm Based on Social Circle Division and Encounter Time Prediction in Opportunistic Network

在机会网络中,节点的行为模式会表现出一定的社交特征,节点往往因相似的移动模式和固定的活动范围,在动态异构网络中表现出一定的集群特性,形成一个个拥有相似社会特征的独特小团体.节点的社会属性表现出长期的稳定性,可以有效应用在路由中.针对这一思想,本文提出了基于社交圈划分和相遇时间预测的机会网络路由算法SCEP(Social Circle division and Encounter time Prediction).该算法关注两个节点形成的直接关系与节点的社会属性特征,定义了基于强社交关系的熟悉集合拓扑,基于熟悉集合的概念以分布式方式开发社区,节点社区的合并受某些规则的约束,并对过时节点进行拓扑剪裁.同时,本文基于节点间相遇的时间间隔序列建模,利用节点间相遇历史数据预测下一次通信的时间.消息的路由通过利用社区、亲密节点集和可预测的通信时间等因素来实现.仿真实验结果表明,与EpSoc,CARA,SAAD,Prophet、NBAPR这5种算法相比,SCEP的性能更好.

In opportunity networks,the behavior patterns of nodes exhibit certain social characteristics,and nodes often show certain clustering characteristics in the network due to similar movement patterns and fixed activity ranges,forming a unique clique with similar social attributes.The social attributes of nodes exhibit long-term stability and can be effectively applied in routing.Aiming at this idea,this paper proposes SCEP,an opportunity network routing algorithm based on social circle division and encounter time prediction.This algorithm emphasizes the direct relationship between two nodes and the social attributes of nodes,defines the familiar set based on the topology of strong social relations,develops the community based on the outward diffusion of the familiar set,interferes with the formation and merging of the community with some constraints,and deletes the obsolete nodes.Also,this paper models the time interval sequence of encounters between nodes and uses the historical data of encounters between nodes to predict the time of the next communication.Messages are routed by leveraging factors such as community,intimacy groups,and predictable communication times.The simulation experimental results show that SCEP performs better than the five algorithms EpSoc,CARA,SAAD,Prophet and NBAPR.