Distributed Topology Manipulation to Control Epidemic Spreading Over Networks

作     者：Xue, Dong Hirche, Sandra 

作者机构：Tech Univ Munich Chair Informat Oriented Control D-80209 Munich Germany 

出 版 物：《IEEE TRANSACTIONS ON SIGNAL PROCESSING》 (IEEE信号处理汇刊)

年 卷 期：2019年第67卷第5期

页      面：1163-1174页

核心收录：

学科分类：0808[工学-电气工程] 08[工学] 

基　　金：German Research Foundation within the Priority Program SPP 1914 "Cyber-Physical Networking" joint Sino-German Research Project "Control and optimization for event-triggered networked autonomous multi-agent systems (COVEMAS)," through the German Research Foundation (DFG) National Science Foundation China 

主　　题：Epidemic spread distributed computation link operation event-based control 

摘      要：This paper considers the problem of network structure manipulation in the absence of a central coordinator that collects network information and makes decisions. The primary focus is on the spectral radius minimization problem by removing/rewiring links to control epidemic spreading over networks. The resulting optimizations are generally combinatorial and NP-hard. The lack of the central base prevents us from solving such problems in a centralized fashion, thus driving the need for distributed computation and collaborative decision-making to support effective topology modification. In this paper, a distributed estimation scheme involving event-based communication and parallel algorithms is developed to enhance network capability against epidemics. Based on this low-complexity estimation algorithm, fully distributed strategies are proposed to enable individuals to sequentially discern the critical in-network contact, and to implement the desired link operation solely. We further extend the acquisitions to a more flexible framework based on a coevolutionary networks picture and discuss the optimality of solutions from an algebraic and topological perspective. Extensive simulation examples are presented to demonstrate the effectiveness of the proposed strategies.