Secure and Sustainable Public Service Delivery with AI, Blockchain and IoT

Abstract

The mix of AI, Blockchain, and IoT looks very promising for making public services in smart cities more secure and reliable. In this work we used a blockchain-based green edge computing dataset and tested anomaly detection on IoT streams using LSTM and GRU autoencoders. Both models gave perfect precision (1.0), but the GRU model performed better overall with higher recall (0.40 vs. 0.30) and F1-score (0.57 vs. 0.46), which shows it can catch more abnormal cases when the clean baseline data is limited. We also checked blockchain validation, that means classification accuracy of predictive models was modest (about 52%), but adding blockchain raised anomaly detection rates from 48.4% (database only) to 52.8%. This shows blockchain does help in tamper resistance. Then we combined AI accuracy, blockchain validation, and latency cost into a composite reliability index.Results showed AI-only had the best mean reliability (0.52), blockchain only was more variable (0.47), while the combined AI–Blockchain score was lower (0.30 mean) but more stable with reduced variance. Hence the study proves AI lowers false alarms, blockchain locks data against tampering, and together they give balanced reliability. Together they offer a solid framework for transparent, tamper-proof, and resilient IoT public services, supporting efforts like Digital India and applications such as smart grids, ration distribution, and city governance.