Intellectual Property India Publishes Patent Application for Machine Learning-Based Predictive Fault Detection And Self-Healing Mechanism For Distributed Computing Systems

MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202641124706 A) filed by Dr. Naga Mallikharjunarao Billa; Mr. Raju Thommandru; Dr. K. Kiran Kumar; Dr. Madhava Rao Maganti; Mrs. Tdn. Pavani; Mrs. P. Surekha; Dr. Challapalli Jhansi Rani; Mr. Podila Purna Chandra Rao; Mr. Pagidi Prem Chand; and Dr. B. Vamsy Krishna on October 18, 2026, for Machine Learning-Based Predictive Fault Detection And Self-Healing Mechanism For Distributed Computing Systems.

Inventors include Dr. Naga Mallikharjunarao Billa; Mr. Raju Thommandru; Dr. K. Kiran Kumar; Dr. Madhava Rao Maganti; Mrs. Tdn. Pavani; Mrs. P. Surekha; Dr. Challapalli Jhansi Rani; Mr. Podila Purna Chandra Rao; Mr. Pagidi Prem Chand; and Dr. B. Vamsy Krishna.

The application for the patent was published on October 30, 2026, under issue no. 44/2026.

Abstract: A machine-learning-based mechanism for predictive fault detection and self-healing in distributed computing systems is disclosed. Telemetry from distributed nodes and services is normalized and aligned to generate an operational state representation from which temporal and cross-component features are derived. A predictive engine processes the features to estimate fault risk and optionally determine a fault category, confidence measure, severity, or prediction horizon. A decision component selects a candidate mitigation using the prediction together with service criticality, dependency topology, resource headroom, action history, and operational constraints. A policy validator checks the candidate mitigation against authorization, safety, redundancy, and impact limits before execution by a self-healing orchestrator and action executor. An outcome monitor verifies post-action health against recovery criteria and a baseline, classifies the result, and provides feedback for baseline adjustment, model calibration, or retraining. The mechanism links prediction, constrained action, and measured recovery in a closed-loop process while supporting monitoring-only, approval-based, and autonomous operating modes. Accompanied Drawing [FIGS. 1-2].

Disclaimer: Curated by HT Syndication.