Approved public publications.

ENFSystems | 2025-10-09 | v1.0.1

Embedded Neural Firmware (ENF) reframes device software as a sealed Neural BIOS- offline-only, immutable, and bound to silicon-and as a framework that specifies, builds, and proves it. ENF compiles a task-specific, quantized model into ROM/Flash and executes bare-metal via a single FSM with static memory (no heap/GC or recursion) and per-state WCET, yielding bit-for-bit reproducibility and tractable verification. Identity and provenance derive from a Physically Unclonable Function (PUF), eliminating stored secrets and gating actuation through measured boot. Powered by harvested energy and supercapacitors under explicit thresholds (V_on/V_safe/V_cut), ENF targets years-long autonomy while collapsing the remote attack surface (no IP stack, OTA, or telemetry). We present ENF's invariants and the ENF Framework-a locked, reproducible toolchain and a Conformance Pack (signed Manifest, firmware/model hashes, PUF-bind record, SBOM, test vectors)-and ground the approach in three cloud-free deployments with a clear threat and limitations posture (recall/replace, dataset governance). Compared with TinyML and Cloud/Edge-AI, ENF offers privacy by architecture and assurance through reproducible builds-evidence you can verify today.

Danish Z. Khan | R3

Domestic water leaks are a rising threat to the financial stability of homeowners and insurers in cold-climate Europe. A press release dated 7 November 2025 reports that burst pipes and leaky fittings generated €4.9 billion in insured losses in Germany during 2024 and that more than half of all building-insurance claims now stem from water damage (Finanztip, 2025). Complementary data from the German Insurance Association show 1.17 million claims in 2021 with an average payout of €3,213 (Maklermagazin, 2022). Existing "smart" shut-off valves tend to be notification-first gadgets that depend on mains electricity, disposable batteries and cloud servers; when storms or freeze events take down power and internet, they leave the home unprotected (Moen Incorporated, 2023). This paper presents ENF AquaFuse / PipeGuard, a patent-informed concept for a batteryless, offline mitigation device that integrates a micro-turbine energy harvester, supercapacitor storage, deterministic embedded neural firmware (ENF) and a mechanical spring-loaded fail-safe valve with manual override. A documented search protocol covering Espacenet, WIPO Patentscope, Google Patents, manufacturer manuals and academic literature found no evidence of any product or patent combining all six resilience criteria (automatic shutoff, self-powered operation, flow-energy harvesting, supercapacitor storage, offline autonomy and mechanical fail-safe) as of 27 December 2025. Hydraulic and electrical calculations show that a 10 F, 5 V supercapacitor bank stores roughly 125 joules-sufficient for sensing, inference and multiple actuation attempts-while a bypass micro-turbine can harvest tens to hundreds of milliwatts without exceeding acceptable pressure-drop limits (Li & Chong, 2019). The architecture accounts for EU safety and hygiene regulations including the Pressure Equipment Directive (PED 2014 /68/EU) (European Parliament & Council of the European Union, 2014), the German UBA evaluation criteria for drinking-water materials and the General Product Safety Regulation (GPSR 2023/988). Comparative pricing suggests that production costs of €45-70 and retail prices of €149-249 would undercut existing mains-powered devices while eliminating maintenance. The ENF AquaFuse concept thus represents a new class of infrastructure-grade leak-prevention hardware; this paper synthesises the patent landscape, presents a worked feasibility example and proposes a validation plan. For full competitive evidence and search protocol details, see Appendices A and C. Keywords: water-leak prevention, energy harvesting, embedded neural firmware, supercapacitor, mechanical fail-safe, EU compliance, cold climate, leak detection, patent landscape, unit economics

Danish Z. Khan

Embedded AI has moved from simple, static firmware into cloud-dependent, update-driven systems that expand attack surface, weaken determinism, and create lifecycle fragility when vendors, networks, or update pipelines fail. This technical note frames that failure mode- energy, security, and privacy limits of the mainstream IoT/LLM/OTA paradigm-and proposes Embedded Neural Firmware (ENF) as an architectural exit. ENF is a firmware-class intelligence stack: a task-specific, quantized neural agent sealed into ROM/flash, executed offline, OS-free, and without OTA, designed to run within harvested-energy constraints. Trust is anchored in hardware (e.g., PUF-rooted identity) rather than remote infrastructure, and behavior is bounded by deterministic control flow and static I/O envelopes. We outline ENF's design commitments, technical grounding (determinism, lifecycle resilience, energy autonomy, security), core contributions, and an optional path toward multi-ENF ecosystems via strictly bounded, non-cloud coordination.

Danish Z. Khan

This technical note surveys the most relevant prior work and standards that shape the design space for Embedded Neural Firmware (ENF). It reviews the TinyML ecosystem and its prevailing assumptions (runtime interpreters, flexible toolchains, and operational updateability), then contrasts these with ENF's sealed, deterministic "compile-and-flash" model. The note summarizes secure-boot and device-identity approaches (including PUF-rooted trust), and examines energy-harvesting constraints that motivate voltage-threshold gating and burst-mode inference. It also outlines formal verification methods for bounded neural networks (e.g., SMT- based and abstract-interpretation approaches) and discusses why ENF-style architectural constraints can make verification more tractable. Finally, it highlights empirical results from adjacent domains that support ENF's core claims: firmware-class inference is feasible, diagnostics can be non-invasive, and strong guarantees require minimizing runtime variability. Overall, the purpose of this note is to position ENF precisely within existing literature-by identifying what exists, what is missing, and what ENF intentionally forbids.

Danish Z. Khan

This technical note specifies a reference architecture for Embedded Neural Firmware (ENF)- a sealed, deterministic embedded intelligence stack designed for MCU-class devices operating without an operating system, cloud dependency, telemetry, or over-the-air updates. The architecture is structured across six layers: (1) a constrained hardware substrate (MCU-scale Flash/SRAM) with hierarchical power domains; (2) an energy-qualified execution model using analog threshold gating, supercapacitor buffering, and interrupt-free finite-state control; (3) firmware-sealed neural inference using statically compiled INT8 models with fixed memory layout, bounded latency, and no dynamic allocation; (4) a security and trust architecture anchored in PUF-derived identity and a ROM-sealed secure boot measurement chain; (5) deterministic safety and fallback behavior based on voltage gating, watchdog deadline enforcement, fail-dormant sink states, and cold-reset atomicity; and (6) a modular multi-node topology supporting deterministic interconnects (e.g., MBus/SPI-class buses) with ROM- encoded addressing, slot-timed signaling, passive reception, and provenance-bound messaging. The note emphasizes auditability, reproducibility, and attack-surface minimization through architectural finality, while explicitly acknowledging tradeoffs: reduced adaptability, no post-deployment patching, and strict task-bounded model scope.

Danish Z. Khan

This technical note formalizes Embedded Neural Firmware (ENF) as a framework-level method for generating deterministic, sealed embedded intelligences from a compact design tuple. It defines ENF as a compile-time system: 𝐸𝑁𝐹(𝑇, 𝑃, 𝑆, 𝐹, 𝐶) → ℒ𝓈ℯ𝒶ℓℯ𝒹 where Task Complexity (T), Power Model (P), Security Level (S), Fallback Architecture (F), and Communication Scope (C) fully specify an agent whose behavior is fixed at manufacture-OS-free, offline by design, telemetry-free, and without OTA updates or dynamic allocation. The note introduces the ENF-Gene as a functional identifier that encodes the five- parameter ontology for cataloging, certification, reproducibility, and long-term auditability. A manifest-driven compiler pipeline is proposed (enf-manifest.yml) that deterministically emits sealed firmware plus cryptographic fingerprints (manifest.hash) and human-legible classification (agent.gene). Security is extended beyond digital integrity into the physical domain via an optional Dual-Gate activation mechanism that requires both PUF-derived identity validation and a factory-paired analog power-signature match prior to execution. The note further defines hybrid neural-formal agents as a practical default architecture, combining bounded symbolic safety envelopes with compact quantized inference for ambiguity handling while preserving analyzable timing, energy, and memory bounds. Finally, the document outlines application classes and a standardization direction for ENF as a taxonomy of permanent, privacy- preserving, single-purpose devices designed to remain trustworthy and reproducible over multi- decade lifecycles.