MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202611091212 A) filed by Malaviya National Institute Of Technology Jaipur on July 27, 2026, for Explainable Spiking Neural Network System For Network Intrusion Forensics.
Inventors include Herschelle Mathew Thomas; Vikash Kumar; and Nand Kumar Jyotish.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: The invention discloses a processor-implemented software system for explainable network intrusion analysis using pcap replay, flow-level feature extraction, spiking neural network inference, and automated forensic evidence generation. The system receives packet capture data or equivalent network traffic records, converts packets into bidirectional flow records using an Argus-based extraction stage, maps the flow information into a trained engineered feature space, and encodes the features into spike trains for processing by a hierarchical spiking neural network. Instead of providing only a class label or confidence score, the system generates synchronized forensic artifacts for selected flows, including Decision Timeline, Neuron Fingerprint, Feature Evidence, and Forensic Report outputs. A working prototype processed a 2 GB UNSW-NB15 pcap, extracted 46,262 flows, produced 4,142 attack decisions, and generated 20 four-module forensic evidence cases with PNG and JSON outputs, thereby supporting explainable software-based cyber forensics, and future live network deployment in practical network security investigation and monitoring.
Disclaimer: Curated by HT Syndication.
