Presentation Information
[B-7-19]Proof-of-Concept Study of a Dual-Blockchain Architecture for Water Treatment Plants
◎△Ryota Sugano1, Shun Fuse1, Masaki Yoshii1, Osamu Mizuno1 (1. Kogakuin University)
Keywords:
blockchain,IT/OT-integrated,storage resource,Hyperledger Fabric,SWaT model
In IT/OT-integrated environments, sensor data and operation logs must be stored in a tamper-resistant manner for cause tracing and auditing. However, storing high-frequency data in blockchain increases ledger size. Our previous study proposed a dual-ledger architecture with a temporary ledger and an aggregation ledger to reduce storage usage. However, it used preformatted data and did not address raw sensor data preprocessing or anomaly detection before blockchain registration.
This paper investigates a dual-ledger architecture with edge computing. Edge devices collect raw sensor data and perform formatting, smoothing, and anomaly detection before sending processed data to the temporary ledger. When an anomaly is detected, the reason and raw data are also recorded. The aggregator generates aggregated data under predefined conditions. The aggregation ledger stores statistical information during normal operation and critical logs when anomalies occur. An application architecture for the Secure Water Treatment (SWaT) model is presented, where one edge device is assigned to each of six stages. Future work will evaluate performance, storage reduction, and anomaly detection validity.
This paper investigates a dual-ledger architecture with edge computing. Edge devices collect raw sensor data and perform formatting, smoothing, and anomaly detection before sending processed data to the temporary ledger. When an anomaly is detected, the reason and raw data are also recorded. The aggregator generates aggregated data under predefined conditions. The aggregation ledger stores statistical information during normal operation and critical logs when anomalies occur. An application architecture for the Secure Water Treatment (SWaT) model is presented, where one edge device is assigned to each of six stages. Future work will evaluate performance, storage reduction, and anomaly detection validity.