BIONIB: Blockchain-based IoT using Novelty Index in Bridge Health Monitoring
Divija Swetha Gadiraju, Ryan McMaster, Saeed Eftekhar Azam, Deepak Khazanchi
TL;DR
This work presents BIONIB, a blockchain-based framework for bridge health monitoring that stores real-time IoT sensor data on the EOSIO blockchain and uses an unsupervised Novelty Index to identify unhealthy bridge conditions. By computing NI at the edge and recording NI-derived transactions on-chain via smart contracts, the approach achieves secure, tamper-resistant data storage and scalable health predictions. The performance analysis demonstrates high throughput and parallel processing capabilities of EOSIO, with a trade-off between latency and resources that remains acceptable for bridge health applications. The results suggest BIONIB can robustly scale to multiple bridges and sensor networks, with future expansion to privacy-focused or private EOSIO deployments for broader infrastructure monitoring.
Abstract
Bridge health monitoring becomes crucial with the deployment of IoT sensors. The challenge lies in securely storing vast amounts of data and extracting useful information to promptly identify unhealthy bridge conditions. To address this challenge, we propose BIONIB, wherein real-time IoT data is stored on the blockchain for monitoring bridges. One of the emerging blockchains, EOSIO is used because of its exceptional scaling capabilities for monitoring the health of bridges. The approach involves collecting data from IoT sensors and using an unsupervised machine learning-based technique called the Novelty Index (NI) to observe meaningful patterns in the data. Smart contracts of EOSIO are used in implementation because of their efficiency, security, and programmability, making them well-suited for handling complex transactions and automating processes within decentralized applications. BIONIB provides secure storage benefits of blockchain, as well as useful predictions based on the NI. Performance analysis uses real-time data collected from IoT sensors at the bridge in healthy and unhealthy states. The data is collected with extensive experimentation with different loads, climatic conditions, and the health of the bridge. The performance of BIONIB under varying numbers of sensors and various numbers of participating blockchain nodes is observed. We observe a tradeoff between throughput, latency, and computational resources. Storage efficiency can be increased by manifolds with a slight increase in latency caused by NI calculation. As latency is not a significant concern in bridge health applications, the results demonstrate that BIONIB has high throughput, parallel processing, and high security while efficiently scaled.
