Evidence

50 km degraded drone mission simulation evidence.

HOLDFAST improves delivery and continuity. Sentry estimates receiver-visible severity and uncertainty. AURM improves trust correctness — it does not change PDR, by design.

Current Evidence State

Simulation-backed architectural prototype.

Current evidence is deterministic and simulation-backed. It compares the same 50 km drone mission, node set and degraded event stream across four configurations: an AODV baseline, HOLDFAST Core alone, HOLDFAST Core with Adapt, and the full stack including Sentry and AURM.

The simulation isolates three separable contributions: delivery and continuity (HOLDFAST), receiver-visible severity and uncertainty (Sentry), and trust correctness (AURM).

The Scenario

50 km degraded drone mission.

Mission

Distributed drone mission across approximately 50 km with relay-dependent C2, packetised video, GNSS/PNT inputs and custody transitions.

Comparison

Same mission, same nodes, same threat environment. AURM/HOLDFAST/Sentry OFF versus ON across four progressive configurations.

Measurement

Mean C2 packet delivery rate, trust-state divergence, custody continuity, operator review signalling and evidence traceability.

HOLDFAST Delivery Findings

HOLDFAST improves C2 delivery.

HOLDFAST Core lifts mean C2 PDR from the AODV baseline through custody, priority and store-forward behaviour. HOLDFAST Adapt adds further gain via evidence-confirmed bounded adaptation when degradation persists long enough to act on.

AODV baseline

0.659

Reactive routing baseline

HOLDFAST Core

0.725

Custody, priority, store-forward gain

HOLDFAST Core + Adapt

0.797

Evidence-confirmed adaptation gain

HOLDFAST Core + Adapt + Sentry + AURM

0.797

Same PDR. Assurance added.

Key message

HOLDFAST improved C2 delivery. AURM did not change PDR, because AURM's role is trust assessment, not packet delivery.

Scenario Deltas

Bounded effectiveness, by scenario.

HOLDFAST performs best under persistent or schedulable degradation. Under transient or ambiguous degradation, Sentry identifies cases where overconfident adaptation should be reduced — by design.

  • Static communications pressure+25.0 pp
  • Packetised video congestion+19.6 pp
  • Relay rejoin stale-command scenario+17.1 pp
  • Combined GNSS / video / C2 degradation+15.0 pp
  • Swept communications pressure+13.8 pp
  • Micro-dwell pressure+5.2 pp
Sentry Uncertainty Findings

Sentry suppresses overconfident adaptation under ambiguous communications evidence.

Sentry estimates receiver-visible severity, uncertainty and evidence quality. When the underlying degraded-state evidence is ambiguous, Sentry restrains the system from adapting as if it had high-confidence information. This is the conservative path: it does not increase PDR on its own, but it reduces the false-confidence failure modes that would otherwise propagate into AURM's trust assessment.

AURM Assurance Findings

Same PDR. Better trust correctness.

Adding AURM on top of HOLDFAST Core + Adapt + Sentry produces the same mean C2 PDR — by design. AURM's contribution is detected in trust outcomes: stale commands prevented from false trust, custody-ambiguous transitions flagged for review, context-invalid commands rejected, and operator-facing assurance signalling on edge cases that would otherwise pass authentication unchallenged.

Relay Custody vs Staleness

Authenticated does not mean current.

In the relay-rejoin scenario (+17.1 pp PDR), HOLDFAST's custody and store-forward behaviour preserves message delivery across the rejoin event. AURM separately determines that some of those delivered commands are stale relative to the current mission state and signals operator review or rejection rather than letting them flow into autonomy policy as fresh instructions.

Interpretation

What this proves. What it does not.

What this proves

  • The Resolvix three-layer architecture is internally consistent and deterministically reproducible in simulation.
  • HOLDFAST measurably improves C2 delivery and continuity under persistent or schedulable degradation.
  • Sentry measurably suppresses overconfident adaptation under ambiguous communications evidence.
  • AURM measurably improves trust correctness, review signalling and false-trust prevention without changing PDR.

What this does not prove

  • Field, RF, GNSS anti-spoof, anti-jam, SDR or waveform validation.
  • Operational deployment, procurement endorsement or combat validation.
  • Specific anti-adversary capability against real-world emitters or threat behaviour.
  • Performance under representative constrained-node hardware. That is the M1 gate.
Next Validation Gate

M1 — live RF / GNSS / HIL evidence.

Current state

Simulation-backed architectural prototype.

Next gate

M1 — live RF/GNSS ingestion, SDR or radio-adjacent evidence, hardware-in-the-loop replay, constrained-node performance, and partner-informed validation.

Not yet claimed

Field validation, RF validation, GNSS anti-spoof validation, anti-jam validation, SDR validation, waveform validation, operational deployment or combat validation.

Claim Boundary

All findings on this page are deterministic simulation evidence from the 50 km drone scenario family. They are not field, RF, GNSS, anti-jam, anti-spoof, SDR or operational validation. The next technical gate is M1: live RF/GNSS ingestion, SDR or radio-adjacent evidence, hardware-in-the-loop replay, constrained-node performance, and partner-informed validation.

Next step

Discuss M1 Validation Partnership.

We welcome discussions with qualified defence, research, autonomy, RF, GNSS, SDR and tactical communications partners able to support the next validation gate: live evidence ingest, hardware-in-the-loop replay and constrained-node testing.