AquaSense-ICS releases leakage-aware water CPS attack benchmark
Original title: AquaSense-ICS: A Hybrid Real-Simulated Cyber-Physical Dataset and Leakage-Aware Benchmark for Smart Water Quality Monitoring and Attack Detection
Under chronological splits, no model beats a constant normal predictor, and unseen-attack AUROC averages 0.64.
AquaSense-ICS is an open hybrid dataset pairing 8.0 hours of Raspberry Pi 5 sensor recordings with a 13-node EPANET-MSX water network. It includes parameterized false-data injection, replay, denial-of-service and drift attacks, sensor faults, and generated sequences with sample-level labels. The authors show random row splits produce a misleading macro-F1 of 0.9998 and propose chronological, held-out-attack evaluation. Data is released under CC BY 4.0 and code under MIT.
- The edge recording has 23,688 records at 1-2 second sampling.
- Random Forest scored macro-F1 0.9998 under the rejected split.
- Best unseen-attack AUROC ranges from 0.35 to 0.93 by type.
- Files are posted at Zenodo DOI 10.5281/zenodo.23390541.
- AI models
- TimeGAN
From the original article
1 min read at arxiv.orgMachine-learning research on water cyber-physical systems (CPS) is constrained by datasets that are either real but chemically poor, or simulated but free of sensing noise and attacks. We present AquaSense-ICS, an open dataset and benchmark that co-registers (i) a 25. 4 h recording window (8.
By Amal Neji · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.