论内容可寻址网络中节点扰动引起的结构变化

Study on Peer Churn in Content-Addressable Network for Distributed Simulations

在内容可寻址网络仿真系统中,节点扰动能够导致区的布局结构发生变化,并且引起整体效率降低.针对这种情况,提出了新的解决方案,即在节点扰动之后的区重组过程中,使用分布式树型数据结构进行优化.仿真结果表明,该解决方案极大地降低了组织过程消耗的网络资源,使系统拥有更高的效率和更好的可扩展性.

A new solution for re-organization of the area was introduced into（ ideally） equally sized rectangular zones upon peers leaving content-addressable network for distributed simulations（ CANS） as a distributed infrastructure to run massive simulations（ for example MMVE games or city traffic simulation）. The peers handle the simulation of zones which is assigned to them. The zones should be split in such a way that there is as little communication between the peers as possible. Because a car or player needs in average the maximum time to cross a zone,a peer to peer tree structure（ CANS tree） was introduced that helps to re-organize the area when peers leave.