Online detection of cascading change-points

TL;DR

This paper introduces an online method for detecting cascading failures in networks by modeling multiple correlated change-points with a diffusion network, using a sequential Shewhart procedure based on likelihood ratios.

Contribution

It presents a novel sequential detection approach for cascading failures using a diffusion network model and addresses computational challenges with unknown parameters.

Findings

Effective detection of cascade failures demonstrated in numerical experiments.

The proposed method handles unknown post-change distributions.

Good performance in real-time failure detection scenarios.

Abstract

We propose an online detection procedure for cascading failures in the network from sequential data, which can be modeled as multiple correlated change-points happening during a short period. We consider a temporal diffusion network model to capture the temporal dynamic structure of multiple change-points and develop a sequential Shewhart procedure based on the generalized likelihood ratio statistics based on the diffusion network model assuming unknown post-change distribution parameters. We also tackle the computational complexity posed by the unknown propagation. Numerical experiments demonstrate the good performance for detecting cascade failures.