MUMBAI, India, Sept. 14 -- Intellectual Property India has published a patent application (202641097190 A) filed by Dr. Neelam Sanjeev Kumar; Srinath S; Yogesh S; and Saran P on August 12, 2026, for Secureslice-Rl: Federated Multi-Agent Rl With Gnn Topology Control For Self-Defending 6g networks.
Inventors include Dr. Neelam Sanjeev Kumar; Srinath S; Yogesh S; and Saran P.
The application for the patent was published on September 11, 2026, under issue no. 37/2026.
Abstract: As 6G networks adopt extreme network slicing to serve heterogeneous services (holographic communication, massive loT, URLLC), each slice becomes an independent attack surface that centralized security systems cannot monitor or defend at scale without violating tenant data privacy. Existing intrusion-detection approaches for 5G/6G rely on centralized machine learning or static threshold rules, which cannot reason about how an attack propagates across the network's topology, cannot coordinate a response across slices without human intervention, and require raw traffic data to leave the edge creating privacy and scalability bottlenecks. SecureSlice-RL addresses this by deploying a federation of reinforcement-learning agents at each slice/edge node. These agents detect threats locally, model the live network as a graph using a Graph Neural Network so the system can reason about how a compromise at one node threatens its neighbors, and coordinate a joint defense quarantining nodes, rerouting slices, or spinning up deception. slices — without any raw data leaving its origin node. A federated aggregation layer shares only model updates across agents, and an explainability layer scores the trustworthiness of each agent's contributions, guarding against poisoned updates. The result is a self- defending 6G fabric that detects, localizes, and autonomously reconfigures around threats in real time.
Disclaimer: Curated by HT Syndication.
