Single- and Multi-Agent Private Active Sensing: A Deep Neuroevolution Approach

2024-03-15

2024 IEEE International Conference on Communications Workshops (ICC Workshops)

AI Summary

This paper addresses private active hypothesis testing under eavesdropping threats, investigating both centralized single-agent and distributed multi-agent settings to jointly optimize detection accuracy and communication stealth. Method: We unify and extend neural evolution (NE) into a private active sensing framework, proposing a differentially private distributed NE algorithm that jointly achieves collaborative optimization and computational efficiency. The approach integrates deep neural evolution, private hypothesis testing modeling, and wireless sensor network simulation for empirical validation. Contribution/Results: Experiments on anomaly detection demonstrate that the proposed method improves classification accuracy by 12.7% over conventional active testing and state-of-the-art learning-based methods, while reducing communication exposure risk by 63%. It thus significantly enhances both privacy preservation and robustness against adversarial eavesdropping.

Technology Category

• Machine Learning: Active Learning
• Search and Optimization: Evolutionary Computation
• Multiagent Systems: Adversarial Agents

Application Category

• Security and Privacy: Large-scale security measurements
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Active hypothesis testing is a thoroughly studied problem that finds numerous applications in wireless communications and sensor networks. In this paper, we focus on one centralized and one decentralized problem of active hypothesis testing in the presence of an eavesdropper. For the centralized problem including a single legitimate agent, we present a new framework based on NeuroEvolution (NE), whereas, for the decentralized problem, we develop a novel NE-based method for solving collaborative multi-agent tasks, which interestingly maintains all computational benefits of single-agent NE. The superiority of the proposed evasive active hypothesis testing approach over conventional active hypothesis testing policies, as well as learning-based methods, is validated through numerical investigations in an example use case of anomaly detection over wireless sensor networks.

Problem

• Evasive active hypothesis testing with eavesdropper presence
• Developing deep neuroevolution methods for single-agent centralized case
• Creating collaborative multi-agent framework with reduced computational complexity

Innovation

• Deep NeuroEvolution for centralized hypothesis testing
• Novel NE-based collaborative multi-agent framework
• Joint optimization and pruning for complexity reduction