AI-Driven Fault Detection and Self-Healing Framework for Resilient Distributed Software Systems in Mission-Critical Application

Fault detection is a crucial component in maintaining the reliability of mission-critical distributed software systems. Traditional fault detection methods rely on rule-based techniques and static thresholding, which suffer from high false-positive rates, delayed fault identification, and limited adaptability. To address these challenges, this study proposes an intelligent fault detection and self-healing framework that integrates anomaly detection, predictive analytics, and reinforcement learning. The model continuously learns from system behavior, proactively identifying and reducing faults, thereby improving system resilience and reducing downtime. The framework was tested using a dataset comprising system logs, network traffic patterns, and performance metrics. Experimental results demonstrated superior fault detection accuracy, reduced Mean Time to Detect (MTTD), and minimized Mean Time to Recover (MTTR) compared to existing methods. The proposed model also showed enhanced scalability and adaptability across diverse distributed environments. This research establishes a robust foundation for autonomous fault detection and recovery in distributed software systems, significantly improving system reliability. Future work will focus on refining learning capabilities, integrating blockchain for secure fault logging, and deploying the model in real-world industrial applications.