Sequential Change Point Detection with FDR Control in Reconfigurable Sensor Networks

2025-04-09

AI Summary

This paper addresses online multiple change-point detection in dynamic reconfigurable sensor networks, where the sensor set evolves over time, complex dependencies exist both across sensors and across time, and change points at individual nodes are independent with abrupt distributional shifts. Method: We introduce e-values—novel sub-uniform statistics—to sequential change-point detection for the first time, proposing an online method that strictly controls the false discovery rate (FDR) at any time step. A dynamic graph model captures evolving inter-sensor dependencies, while an adaptive scheduling mechanism handles network reconfiguration, relaxing the conventional static independence assumption. Contribution/Results: The method provides a theoretical guarantee that FDR remains ≤ α at all times. Experiments demonstrate that, under identical FDR constraints, it achieves significantly lower false negative rates (FNR) and average detection delay compared to state-of-the-art approaches, while maintaining robustness and real-time performance.

Technology Category

• Constraint Satisfaction and Optimization: Distributed CSP/Optimization
• Planning, Routing, and Scheduling: Scheduling under Uncertainty
• Reasoning under Uncertainty: Sequential Decision Making

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

This paper investigates sequential change-point detection in reconfigurable sensor networks. In this problem, data from multiple sensors are observed sequentially. Each sensor can have a unique change point, and the data distribution differs before and after the change. We aim to detect these changes as quickly as possible once they have occurred while controlling the false discovery rate at all times. Our setting is more realistic than traditional settings in that (1) the set of active sensors - i.e., those from which data can be collected - can change over time through the deactivation of existing sensors and the addition of new sensors, and (2) dependencies can occur both between sensors and across time points. We propose powerful e-value-based detection procedures that control the false discovery rate uniformly over time. Numerical experiments demonstrate that, with the same false discovery rate target, our procedures achieve superior performance compared to existing methods, exhibiting lower false non-discovery rates and reduced detection delays.

Problem

• Detect sequential change points in reconfigurable sensor networks
• Control false discovery rate during sensor data monitoring
• Handle dynamic sensor activation and dependencies effectively

Innovation

• e-value-based detection procedures for FDR control
• handles dynamic sensor activation and deactivation
• improves detection speed and reduces false negatives