A Solution for Recovering Network Topology with Missing Links using Sparse Modeling

Abstract

In recent years, sparse modeling, which is a statistical approach, has been applied to many practical problems mostly in the fields of signal processing and image processing, and a dictionary construction method and a sparse representation for network topology with sparse modeling have been proposed in the field of information networking. We believe that a dictionary for network topologies can be utilized for various purposes. In this paper, we investigate how the network topology with missing links can be recovered using a dictionary for network topologies constructed with sparse modeling. Specifically, we propose a method called TRSM (Topology Recovery with Sparse Modeling) that recovers missing links using a dictionary constructed from many teaching network topologies using the overcomplete dictionary construction algorithm called the K-SVD algorithm. Furthermore, through experiments, we investigate how accurately the randomly deleted links from a network can be recovered with TRSM.