All defense POCs
POC 19CCompressionSmall build (single system)Wave 2POC
PrecisionEdgePrecisionCommsPrecisionSWaP

PrecisionEdge — Tactical Link Compression

Semantic precision reduction on tactical telemetry — 3–6× bandwidth gain with explicit per-value error bounds. Mixed Precision engine on Link-16 class payloads.

100%
Real-data survival (4 domains)
6.5×
vs zstd (enterprise suite)
89.8×
Telemetry streams (headline bench)

The scenario

Set the picture

A tactical datalink carries fused tracks, sensor telemetry, and status words at fixed bandwidth. gzip on float64 buys little on numerical payloads; float16 throws away the error bound.

Programs need more reporting density on the same link budget — with an auditable per-value error envelope on the wire.

What it costs today

General compressors (zstd, LZ4) achieve ~1.0–1.5× on numerical columns.

Ad hoc float16 downcast loses audit-quality error documentation.

What changes with SolvNum

Mixed Precision profiles each field, reduces semantic precision, then compresses — with explicit relative-error bounds per value at chosen k.

Csemantic compression with bound

Real-data validation: Open-Meteo climate, LBNL PMU, NOAA DSCOVR satellite, OpenSky ADS-B tactical — 100% survival across Wave 1 sweep. Tactical streams up to 89.8× vs 14× best general compressor in the headline benchmark.

Measurable outcome

What we'll claim — and how it survives review

Each line below maps to a captured number in the demo section. Every number is reproducible from the SolvNum validation suite.

  • 100% compression survival on 4 real-world domains (Aegis Wave 1).
  • 6.5× overall advantage vs zstd on enterprise numerical suite; 89.8× on telemetry streams in headline benchmark.
  • Recorded Link-16 / STANAG tactical messages: POC — synthetic field generator validated (G-012).
  • PrecisionSWaP embedded flash/RAM: ROADMAP measurement on target HW (G-014).

The demo

What was tested. How. What the script printed.

Wave 1 compression sweep (Aegis 8244–8256): semantic vs naive vs lossless vs uniform × error bounds × seeds on synthetic + real datasets including OpenSky ADS-B tactical.

Illustration

In-browser diagram of what the demo proves. The numbers underneath are the captured demo output.

Required bandwidth vs. link budget

link budget 2.50 Mbps

Evidence pointers

Where the claims live in the repo

These are the files a reviewer should run, read, or grep to re-derive every number on this page.

  • mixed_precision/portfolio/precision-edge.md
  • mixed_precision/portfolio/overview.md
  • solv/17-defense-validation-gaps.md (G-012, G-014)
  • Aegis jobs 8106–8115, 8244–8256

Want to see this in your environment?

Brief us on a program where this POC matters.

ITAR-aware. Air-gapped delivery available. Every claim above traces back to a script in the public repo.

Brief us