Blockchain Based Firewall System for The Banks: A Hybrid AI-Driven Architecture for Immutable Logging and Adaptive Defense

The increasing adoption of digital banking has significantly expanded the attack surface for cyber threats, including zero-day attacks, ransomware, insider threats, and advanced persistent attacks. Traditional firewall systems, which rely on static rules and centralized log management, are often unable to detect evolving attack patterns or ensure the integrity of security logs. This paper proposes TrustShield, a blockchain-assisted intelligent firewall framework that combines Artificial Intelligence (AI), Federated Learning (FL), Explainable AI (XAI), and permissioned blockchain technology to strengthen banking network security. The framework employs Random Forest, XGBoost, and a Transformer-based intrusion detection model to analyze network traffic and identify malicious activities in real time. Federated Learning enables multiple banks to collaboratively improve detection models without sharing sensitive customer data, while SHAP-based Explainable AI provides transparent reasoning for each security decision. Blockchain and smart contracts ensure tamper-proof event logging, secure firewall policy management, and regulatory compliance. Experimental evaluation indicates improved threat detection accuracy, lower false-positive rates, reduced response latency, and enhanced auditability. The proposed framework offers a scalable, privacy-preserving, and trustworthy cybersecurity solution for modern banking infrastructures.