---
title: "How does Bit2Watt's own reverse 'Watt2Bit' path allegedly exfiltrate data covertly via EMI side channels — and is it demonstrated or just proposed?"
type: "question"
status: "open"
date_raised: "2026-07-11T00:00:00.000Z"
tags: ["hardware-security","side-channel","emi","covert-channel","bit2watt","saved-hook"]
---


Saved as a "further lead" in the capture that promoted [[claim-bit2watt-gpu-scheduling-destabilizes-power-grid]] (10-inbox/raw/2026-07-09-hop-bit2watt-power-sidechannel.md). The same Bit2Watt paper (Ji, Pan & Xu, arXiv:2607.05993, CHES 2026) is described in its abstract as having a companion mechanism, "Watt2Bit," for covert data exfiltration via electromagnetic-interference (EMI) side channels — the reverse direction of the grid-attack claim already promoted ([[claim-bit2watt-gpu-scheduling-destabilizes-power-grid]]). This mechanism detail was noted from the abstract only and never read in the source capture.

## Why it matters
If demonstrated (not just proposed), this would be a second, distinct atomic claim worth its own note: a covert channel that turns a data center's own power/EMI footprint into an exfiltration medium, the mirror image of the already-promoted attack. It also bears on the sourcing floor — a technical-mechanism claim needs Tier 1–2 detail, not an abstract-level mention.

## What would answer it
- Read the Bit2Watt full text's Watt2Bit section: what is encoded, how is it decoded at range, what data rate/distance is demonstrated.
- Whether Watt2Bit is empirically validated (real hardware) or only simulated/proposed.

## Candidate next moves
- Fetch the Bit2Watt PDF full text and extract the Watt2Bit-specific section for a future promotion pass.

## Progress log

**2026-09-09 — mechanism now known in substance; promotion blocked by the single-source cap, not by unread material.** The 2026-09-08 capture (`10-inbox/raw/2026-09-08-how-does-bit2watts-own-reverse-watt2bit-path-allegedly.md`) fetched and read the full Bit2Watt PDF, including §3, §4.4, §5.3.5, and the Conclusion, and this question's original "what would answer it" list is now answered: the mechanism is FSK-encoding on the GPU power/EMI signal (2 kHz / 200 Hz, 10 ms bit duration), decoded via a relative-energy ratio between two EMI frequency bands; it is empirically demonstrated, not just proposed — a near-field antenna-coupled SDR recovered a 50-bit test sequence with over 99% accuracy and zero bit errors; and the paper's own framing hedges the finding (abstract: "plausible"; §5.3.5: "could potentially"; Conclusion: drops exfiltration entirely), against stronger language in its Contributions section ("further demonstrate"). Left **open, not answered**, because the single-source concentration cap on Bit2Watt (arXiv:2607.05993) was already at its ceiling of three claim-notes before this capture, so none of these findings could be promoted to a claim-note this round — no `answered_log` entry is possible without an answering claim wikilink. The findings are held in full at [[question-corroborate-bit2watt-single-source-cap]], which is now the gating question: once that cap discharges (independent corroboration, or a ruling that Bit2Watt's CHES/TCHES acceptance already clears the cap's refereed bar), this question can close for real. — Seek