Resilience analysis for reconfigurable network with cascading failures

考虑级联失效的可重构网络抗毁性研究

¹ School of Mechanical Engineering, Northwestern Polytechnical University, Xi’an 710072, China
² Laboratory of Science and Technology on Integrated Logistics Support, National University of Defense Technology, Changsha 410073, China

Received: 27 October 2020

Abstract

A network is a topology composed of nodes and the edges of connecting nodes. The nodes interact with each other through edges and realize the function of network. The break of one or several nodes in the network will cause the failure of other nodes through the connection relationship between nodes and this phenomenon is known as cascading failure. The function of large network depends on network structure. When some nodes or edges fail, the network can restore and maintain certain functions by means of reconstruction and load distribution. Based on the cluster of unmanned aerial vehicle (UAV) and communication network, the cascading failure and reconfigurable features are considered in this study. By introducing the limitation of the maximum number of connections of nodes, a multi-state cascading failure rule based on information load distribution and a dynamic network reconnection rule based on probability are proposed, and the influence of the spread and propagation of cascading failure behavior on the invulnerability of the network are studied. Through simulation experiment design and analysis, several feasible suggestions to improve the invulnerability of UAV clusters are concluded.

摘要

网络是由节点与连接节点的边构成的拓扑结构；节点之间通过边进行信息交互并实现网络的功能。网络中一个或多个节点发生的故障，会通过节点之间的连接关系引起其他节点发生故障，这种现象称为级联失效。网络的功能依赖于网络结构，当部分节点或者边发生故障时，网络往往能通过重构和载荷分配等方式恢复和保持一定的功能。以无人机集群和通信网络为研究背景，考虑了无人机集群的级联失效行为和可重构特点。通过引入节点最大连接数限制，提出了一种基于信息载荷分配的多状态级联失效规则与基于概率的动态网络重连规则，研究了级联失效行为的扩散和传播对网络抗毁性的影响。通过仿真实验设计及分析，提出几点提升集群抗毁性的可行建议。