MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202641101351 A) filed by G Pullaiah College Of Engineering And Technology on August 21, 2026, for Graph-Based Anomaly Detection In Real-Time Streaming Data And Scalability Optimizations.
Inventors include Dr. M. Sri Lakshmi; Ameena Yasmeen; Dr C. Maddilety; Bestha Chandrakala; D. Navya Devi; U Prem Sagar; K Srinivasulu Achari; and B. Thulasi.
The application for the patent was published on August 28, 2026, under issue no. 35/2026.
Abstract: A graph-based anomaly detection framework is disclosed for identifying unusual structural and temporal patterns in real-time streaming data. The system constructs and incrementally maintains a dynamic graph from incoming edge streams, employing multi- resolution sketching structures combined with adaptive density estimation to flag both individual edge anomalies and emerging dense subgraphs. Scalability is achieved through hierarchical partitioning of the graph state, selective update propagation, and memory- bounded approximate counting mechanisms that avoid full graph materialization. Experimental evaluations on high-volume network traffic streams demonstrate improved detection accuracy relative to constant-memory baselines while sustaining processing rates suitable for continuous operation. The framework supports deployment in environments requiring immediate response to suspicious connectivity patterns without requiring offline batch recomputation.
Disclaimer: Curated by HT Syndication.
