MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202631124358 A) filed by C. V. Raman Global University on October 16, 2026, for A Hybrid Fuzzy-Reinforcement Learning Framework For Dynamic Trust Evaluation In Zero Trust Architectures.
Inventors include Ashutosh Soni; Surendra Kumar Nanda; Jayanti Rout; Mrutyunjaya Sathua; Swatisipra Das; Rojalina Priyadarshini; Adyasha Rath; and Ganapati Panda.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: An adaptive trust evaluation system and method for Zero Trust security architectures is disclosed. The invention resides within a Policy Decision Point and computes a continuous trust score for each access request or active session. Real-time input parameters, including device posture, network risk and behavioural anomaly indicators, are converted into degrees of membership across linguistic categories by a fuzzy inference engine. A rule base is evaluated to produce an initial trust level, which is defuzzified into a crisp numerical score. A reinforcement learning agent observes the outcomes of enforced access decisions, such as breaches or correctly restricted access, and adjusts the weights assigned to each input parameter accordingly. The weighted, adjusted score is mapped to an allow, challenge or deny decision at a policy enforcement point. Because the reward signal is derived from actual security outcomes, the system improves its accuracy over time without manual reconfiguration or full retraining. Fuzzy rules provide human-readable reasoning traces for each decision, supporting audit and compliance requirements. The framework is domain- agnostic and can be deployed in cloud, financial and smart city environments by adapting only the rule base and reward configuration. The method is implemented in software on standard computing infrastructure and performs incremental computation suitable for real-time enforcement.
Disclaimer: Curated by HT Syndication.
