为了尽可能地在本地访问网络资源,从而缩短网络访问时间,提出了一种位置感知的分布式生成树LDST模型,使得地理上邻近的节点被分在逻辑上靠近的组中。通过采用底层网络中节点间跳数或消息延迟作为覆盖网络中两个节点间的距离,制定了节点在LDST模型中代表元和邻近组的选取规则,给出了建立无标杆的、分层的、位置感知的覆盖网络的构造算法。数学分析和仿真结果表明,LDST模型具有小世界特性,节点加入算法具有对数时间复杂度,LDST模型具有较好的易扩展性和应用价值。
This paper proposed a locality-aware distributed spanning tree(LDST) model for trying to access the network resources localized and reduce the latency,by organizing adjacent nodes in geographical into close groups in logical.Firstly,defined the distance of two nodes in the overlay network with the hops or the latency between two nodes in the underlying network.Next,presented the rules how to select the node as the representative and the group as the nearest group in the LDST model,and showed the arrival algorithm to construct a non-landmark,hierarchy and locality-aware overlay network.Finally,the mathematic analysis and simulation results show that the LDST model is a small-world,the time complexity of the arrival algorithm is logarithmic and the scalable LDST model has a good application value.