Anomaly Detection In Blockchain Networks In Upi Transactions

Abstract

Blockchain Technology Has Revolutionized Distributed Systems Through Its Decentralized, Immutable, And Transparent Architecture. However, The Increasing Adoption Of Blockchain Networks Has Attracted Malicious Actors Exploiting Vulnerabilities For Fraud, Money Laundering, And Other Illicit Activities. This Paper Presents A Comprehensive Machine Learning-based Framework For Detecting Anomalies Across Multiple Layers Of Blockchain Architecture. We Propose A Multi-layered Detection System That Integrates Supervised, Unsupervised, And Deep Learning Techniques To Identify Suspicious Patterns In Transaction Flows, Smart Contract Execution, And Network Behavior. Our Evaluation On Bitcoin And Ethereum Datasets Demonstrates 94.7% Detection Accuracy With A False Positive Rate Of 2.3%. The Proposed System Addresses Key Challenges, Including Limited Labeled Data, Real-time Processing Requirements, And Privacy Preservation Through Federated Learning Integration.

Keywords