ENF - Embedded Neural Firmware (v1.0.1)

A sealed, offline "Neural BIOS" for batteryless devices.

Key points

• No cloud, no OTA, no telemetry
• Sealed image + PUF identity (keys derived, not stored)
• Static memory only; fixed tensor shapes; no heap/GC/recursion
• Deterministic timing (WCET) with a single FSM: Wake -> Sense -> Infer -> Actuate -> Sleep
• Energy thresholds with fail-closed brown-out (V_on / V_safe / V_cut)

Downloads

• Whitepaper PDF (repo): ENF-Whitepaper-v1.0.1-2025-10-09.pdf
• Zenodo DOI: https://doi.org/10.5281/zenodo.17298403
• ResearchGate: https://www.researchgate.net/publication/396325650_Embedded_Neural_Firmware_ENF_A_Deterministic_Offline_Neural_BIOS_for_Batteryless_Devices

What is ENF

ENF is a strict pattern for tiny neural firmware: the model is burned into memory and never changes; there is no network stack; timing and memory are fixed and testable; each device proves identity via a silicon PUF; execution only runs when harvested voltage is safe.

Files

• ENF-Whitepaper-v1.0.1-2025-10-09.pdf
• (optional, later) conformance/ — manifest.yml, boot_measurement.json, wcet.csv, energy_thresholds.txt

How to cite

Khan, D. Z. (2025). Embedded Neural Firmware (ENF): A Deterministic, Offline "Neural BIOS" for Batteryless Devices (v1.0.1). Zenodo. https://doi.org/10.5281/zenodo.17298403

License

Paper: CC BY 4.0. See LICENSE.

Contact

galaxyideator@gmail.com • Neumark, Germany

ENF and ENF Manifest are unregistered trademarks of Danish Z. Khan.