MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202641083844 A) filed by M. V. Karthikeya; Lakshya Swaroop G; Pranshu Parth; and D. Nanda Kishore on July 08, 2026, for Aegisai: Offline Biometric Authentication System With Neural Face Recognition And Multi-Stage Livene.
Inventors include M. V. Karthikeya; Lakshya Swaroop G; Pranshu Parth; and D. Nanda Kishore.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: This invention discloses an edge-computed biometric authentication system that operates entirely within a local hardware boundary and produces three distinct measurable technical effects:-(T) elimination of replay-attack vulnerability through hardware-tethered entropy-seeded randomized liveness challenge generation, in which a processor samples physical read noise and shot noise from an optical sensor's photodetector array as the primary entropy source to generate per-session continuous-valued randomized geometric challenge target sub-ranges and presentation order, rendering it physically impossible for any prior-session recording to satisfy a subsequent session's independently-derived parameters; (2) measurable stabilization of authentication decisions on CPU-only hardware through a closed-form, non-learned illumination-stability and structural-deformation-variance weighted multi-frame temporal arbitration computation implementing a three-zone decision architecture—confirmation above approximately 0.78, continuation between approximately 0.58 and 0.78, approximately 0.58—producing reduced false-rejection illumination and intermittent frame capture without GPU classifier parameters; and (3) guaranteed single-delivery through write-ahead-logged transactional persistence with reentrant mutualexclusion locking, exponential backoff retry, and atomic synchronization-flag update conditioned on explicit upstream confirmation, eliminating both data loss and duplication under arbitrary network interruption. The complete authentication pipeline operates within a closed local hardware boundary with no network dependency during active authentication. Each claimed combination is inextricably tied to the specifically-recited hardware structures of the local computing device and produces a specific technical advancement over prior art.
Disclaimer: Curated by HT Syndication.
