A fast algorithm for diffusion source localization in large-scale complex networks

Abstract

The identification of the origin of diffusion processes in complex networks is a subject of significant interest across numerous interdisciplinary fields. One approach to solving this issue involves the placement of a few observer nodes within the network and the estimation of the unknown source through the utilization of information gathered by these observer nodes. However, this approach presents certain drawbacks, particularly with regard to computational complexity. To address this limitation, this study introduces an innovative Hill-Climbing algorithm designed to efficiently identify diffusion sources within large-scale complex networks. Our approach, the Local Search Hill Climbing (LSHC) method, transforms the source localization problem into an optimization task, utilizing strategically deployed observer nodes. Experiments conducted on both random and scale-free network models demonstrate that our method significantly reduces computational time while maintaining high accuracy in pinpointing the diffusion source. This approach offers a substantial improvement over traditional methods and holds considerable promise for practical applications in network science.