MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202621085230 A) filed by Ms. Monika Agrawal; Mrs. G. Chandrika; Dr. P. Subhashini; Mr. B Janu Naik; Dr. Tatiraju. V. Rajani Kanth; Mr. S Niresh Kumar; and Dr. C. R. Rene Robin on July 11, 2026, for System And Method For Confidence-Gated Adaptive Cloud Resource Allocation Using Reinforcement Learning And Multi-Layer Telemetry Feedback.
Inventors include Ms. Monika Agrawal; Mrs. G. Chandrika; Dr. P. Subhashini; Mr. B Janu Naik; Dr. Tatiraju. V. Rajani Kanth; Mr. S Niresh Kumar; and Dr. C. R. Rene Robin.
The application for the patent was published on August 28, 2026, under issue no. 35/2026.
Abstract: A system and method for confidence-gated adaptive allocation of cloud computing resources are disclosed. Multi-layer telemetry is acquired from workloads, virtualisation components and physical computing nodes and is processed to generate a multidimensional resource state. A dynamic infrastructure dependency graph represents workload, computing, network, storage, accelerator, power, thermal and failure-domain relationships. A reinforcement learning engine generates candidate resource allocation actions, and a simulation engine predicts a post-action state for each candidate action. Candidate actions violating service-level, hardware, thermal, power, isolation or redundancy constraints are rejected. A confidence gating engine selects an approved action according to action confidence and predicted technical benefit. The approved action is decomposed into reversible micro-allocation operations. Post- actuation telemetry is compared with the predicted state after each operation to continue, modify, suspend or reverse resource changes. The system reduces allocation oscillation, migration overhead, energy consumption, thermal imbalance and service-level violations while improving utilisation and operational stability of heterogeneous cloud infrastructure. Accompanied Drawing [FIGS. 1-2]
Disclaimer: Curated by HT Syndication.
