聚类的组播组分配策略在DDM中的实现

A Cluster- Based Multicast Group Assignment of DDM

DDM(DataDistributionManagement)服务是高层体系结构HLA(HighLevelArchitecture)中定义的六大服务之一,它主要目的是通过减少网络间不相关数据的分发,提高网络的利用效率。一种常用的DDM优化方法是使用组播,组播是一种非常适合一对多的通信方式,它能够将更新数据发送到一个指定的定购者子集里。传统的DDM组播使用固定网格分配算法,在资源使用和数据发送方面存在一定的问题,不适合超大规模的仿真。本文在固定网格算法基础上提出了一种基于聚类的组播组分配算法,通过聚类,将固定网格的组播组进行合并,减少组播组的使用数量,提高网络资源的利用率,最后从实验结果上分析了聚类组播算法的优缺点。

Data Distribunon Management is one of the six services provided by High Level Arctutecture （HLA）.It aims to make network communication more efficient by reducing irrelevant data between federates.A common optimization in DDM is the using of multicast, it is a befitting transmission way to deliver data from single-source to muldple-destination.A tradition multicast group assignment of DDM based on fixed-grids,but it is not adaptive for large-scale distributed simulations at data distributing and resource utilization. In this paper, we propose a cluster-based multicast group assignment.The basic idea of clustering is to combine multiple publication regions into a cluster.And Last.we analysis the advantage and disadvantage of this algorithm based on theory and experiment.