MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202641091902 A) filed by J R Jayasakthi; and Dr M Meenakshi on July 29, 2026, for Hardware Level Security Via Rf Fingerprints: A Defense Against Spoofing And Unauthorised Access.
Inventors include J R Jayasakthi; and Dr M Meenakshi.
The application for the patent was published on August 07, 2026, under issue no. 32/2026.
Abstract: A system and method for hardware-level security using radio frequency (RI') fingerprints isolates device-under-test (DuT) specific hardware impairments by mathematically cancelling reference transceiver effects. The system comprises a first software defined radio (SDR) (102) as a DuT, a second SDR (104) as a reference transceiver/authenticating device, a synchronization module (105), and a processing unit (106). Four measurement configurations are performed: DuT loopback, reference loopback, DuT-to-reference transmission, and reference-to-DuT transmission. A mathematical formulation computes isolated DuT-only I/Q components by combining loopback and cross-measurement differences using a product-over-division cancellation. A hybrid deep learning classifier (108), comprising a 1 D convolutional neural network and a statistical feature branch, classifies devices based on the isolated fingerprints. Three operational modes address: (i) programmable DuT with known standard, (ii) non-programmable DuT with known preamble (eg., Zigbee), and (iii) unknown DuT with unknown standard using blind cyclostationary estimation. Loopback validation achieves Bit Error Rate (BER) = 0, and two-SDR synchronization at 2.4 GHz yields greater than 97% classification accuracy for five PlutoSDR devices, greater than 98% for Zigbee devices, and greater than 95% for unknown transmitters. The invention enables receiver-independent physical-layer authentication for IoT and wireless networks.
Disclaimer: Curated by HT Syndication.
