Disconnected-Comms Autonomous Handoff
An hour underwater, a day on the lunar surface, a sortie deep in denied airspace — handoff state replays bit-identically on the parent platform, in the comms window you actually have.
- 832 bits
- Compressed handoff packet (0.10 KiB)
- 1.5 Mbit
- Comms budget (99.9% headroom unused)
- ✓
- Continuation hash match (parent = UUV)
- 1.0000×
- Observed per-tick excursion (bound 2.4623×)
The scenario
Set the picture
A UUV operates beyond reliable comms for hours to days. Periodically it surfaces or enters an acoustic-modem comms window and must hand off its accumulated state to a parent platform — a surface ship, a buoy, a relay submarine. The parent must be able to continue the mission from that state, on different hardware, as if it had been the controller all along.
The same constraint binds every disconnected-autonomy use case: lunar / cislunar surface mobility with hours-to-days of round-trip light-speed delay, contested-ISR penetrating UAS in deep denied airspace, submarine periscope-depth burst comms windows, special-operations dismounted teams operating EMCON, satellite-constellation autonomous handoff between operations centers.
What it costs today
Disconnected autonomy lives with three permanent costs. Lossy partial-state download — handoff is whatever fits in the comms window; subtle numerical state divergence between the disconnected platform and the parent is not detected and silently degrades mission continuation.
Bandwidth-vs-fidelity argument — every program lives the same conversation: 'we can downlink X kbps of the state, what do we cut?' The cut is always lossy in ways the operations team cannot quantify. No reconciliation hash — when the parent picks up the mission, there is no way to confirm 'the state I received matches the state the disconnected platform is carrying.' Trust is contractual, not cryptographic.
Programs hesitate to commit to multi-day disconnected autonomy because they cannot verify the asset's state without a comms-rich window. When something goes sideways during the disconnected interval, the post-mission record is whatever the platform recorded onboard — and that record is not identical to what the parent's continuation produced.
What changes with SolvNum
Three capabilities, each carrying part of the disconnected-handoff burden.
The handoff state replays bit-identically on the parent's hardware. The parent is not approximating — it is computing the same numbers the disconnected platform was computing. Any cross-check (compare-hash on a known sub-state, replay a recent decision) is binary: match or no-match.
Per-tick state evolution is bounded, so a 'key-frame plus deltas' transmission scheme has a provable maximum drift between transmitted updates. The parent can re-derive the missing in-between trajectory from the excursion-limit property and verify the bit-identical hand-back hash. This is the trick that makes narrow comms windows usable.
The state itself is compressed at a known error bound, fitting into narrow comms windows — acoustic modem on a UUV (kilobits, not megabits), VLF burst on a submarine, lunar SATCOM with seconds of bandwidth budget per pass. The error bound is on the wire, not in a separate document.
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.
- Multi-hour autonomous gaps reconcilable to bit-identity on reconnection — single hash compare on the handoff artifact.
- Explicit bandwidth-vs-fidelity budget per channel, set by k parameter and key-frame interval, documented in interface control documents.
- Provable maximum drift between transmitted updates, derivable from the excursion-limit property — not a statistical estimate.
- Adverse-event reconstruction across the disconnected interval is bit-faithful within the documented error envelope.
- Long-duration disconnected autonomy programs (multi-day UUV, weeks-long lunar surface, sortie-length deep-ISR UAS) become committable because verification is mathematical.
The demo
What was tested. How. What the script printed.
4-hour simulated UUV mission (14,400 state ticks of 16-coupled-oscillator autonomy state) with a 5-minute acoustic-modem comms window at 5,000 bps (1.5 Mbit total budget). The UUV transmits its compressed handoff package within the window. The parent platform — different hardware, different operating environment — receives the handoff and continues the mission for 1 additional hour of simulated time.
Two stacks compared on the same scenario. Verified: budget fit (C), bit-identical continuation (D), and per-tick bounded evolution (B). The parent's onward continuation is then hash-compared to the UUV's self-recorded continuation.
Illustration
In-browser diagram of what the demo proves. The numbers underneath are the captured demo output.
Mission timeline
Handoff packet vs. comms budget
832 bits / 1,500,000 bits (0.06%)✓ Continuation hash match — parent's mission state and UUV's self-recorded continuation produce the same SHA-256 after 1 additional hour of independent execution on different hardware.
32476fa95ee28e8338ccc88f752837456bf3e7f55f0ff688d8ac952762b899b4
Captured demo output
The numbers the script actually printed.
| Phase | Value |
|---|---|
| Disconnected interval | 4.0 hours (14,400 state ticks) |
| Comms window | 300 s @ 5,000 bps |
| Handoff budget | 1,500,000 bits (183.1 KiB) |
| Compressed handoff packet | 832 bits (0.10 KiB) |
| Budget headroom | 1,499,168 bits (99.9%) |
| Continuation duration | 1.0 hour (3,600 ticks) |
| Per-tick excursion observed | 1.0000× (bound 2.4623×) |
| Continuation hash (parent) | 32476fa95ee2… |
| Continuation hash (UUV self-recorded) | 32476fa95ee2… |
| Match | ✓ |
Disconnected-Comms Handoff Reconciliation Attestation
- Disconnected duration
- 4.0 hours
- Handoff packet size
- 832 bits (104.0 bytes)
- Comms budget
- 1,500,000 bits
- Per-tick bound
- ≤ 2.4623× per-step excursion
- Continuation duration
- 1.0 hour
- Continuation hash match
- True
- Continuation hash
- 32476fa95ee28e8338ccc88f752837456bf3e7f55f0ff688d8ac952762b899b4
Composes with
Where this POC sits in the substrate
Every POC reinforces — and is reinforced by — others. Click through to see how each piece locks into the larger picture.
Mission Rehearsal Parity
Mission Rehearsal Parity is the foundation determinism property.
Provable Effector Slew Rate
Effector Slew Rate is the per-effector application of the same excursion-limit primitive.
Bandwidth-Bounded Tactical Telemetry Compression
Telemetry Compression is the broader application of the compression primitive.
Cross-Platform Mission-Data-Recorder Compression
MDR Compression is the archival counterpart to the live-handoff use case here.
JADC2 Reference Compute Substrate
JADC2 Substrate integrates this with scale-based classification (R) for the full four-capability story.
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.
- SolvNum cross-platform determinism verification (x86, ARM, WASM)
- SolvNum validation suite — excursion-limit verification
- SolvNum streaming-compression demo, benchmark suite — compression primitive + competitive benchmark
- SolvNum benchmark suite — trajectory stability-bounded evolution demo
- SolvNum cross-platform attestation benchmark
Previous · POC 09
Multi-Platform Drone Swarm with Provable Safety
Next · POC 11
Cross-Platform Engagement Replay for Review
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.