DOMAIN — COUNTER-UxS OPERATIONS

c-UxS COA Engine

AI pipeline for unmanned systems counter-operations · FPV · UAS · SWARM · UGV · USV · DE · Validated June 2026

OVERVIEW

Counter-UxS as a Distinct Warfighting Domain

The proliferation of low-cost unmanned systems — FPV kamikaze drones, autonomous swarms, loitering munitions, and UGV/USV teaming — has created a warfighting domain that exceeds the response bandwidth of human-only planning. ConclAive's c-UxS engine applies the same adversarial COA architecture used for conventional land, naval, and air operations to this high-tempo, sensor-rich, multi-UxS environment.

The system ingests a live force disposition, runs a full threat assessment, generates three scored courses of action, and delivers a commander recommendation — in under 90 seconds, fully offline, on hardware ranging from a tactical tablet to an edge server.

17k+ Simulation Runs

0 Pipeline Failures

92.4% Intent Accuracy

7.6s Median Latency

100% Positive Security Gain

3 COAs per Run

SUPPORTED UNIT TYPES

12 c-UxS Unit Classifications

The engine models both unmanned threat vectors and counter-UxS response assets, enabling full blue-red force pairing and mission-specific COA generation.

⬡ FPV First-person view kamikaze drone — high attrition threat, low cost, difficult to intercept at scale

△ UAS Fixed/rotary unmanned aircraft — ISR, strike, relay, and SIGINT roles

⊛ SWARM Coordinated multi-agent attack — saturates point defenses, requires distributed counter-response

▣ UGV Unmanned ground vehicle — breaching, logistics, remote fire support

◇ USV Unmanned surface vessel — littoral, riverine, and harbor threat vector

⊕ SHORAD Short-range air defense — kinetic counter to UAS/FPV threats

⚡ DE Directed energy — high-repetition counter-UxS, low cost per kill

⊗ C_SUAS Counter-small UAS — electronic defeat, nets, kinetic, multi-layer response

⊘ EW Electronic warfare — GPS/RF jamming, spoofing, link disruption

◉ RECON ISR platform — persistent overwatch, target cueing, BDA

□ CARGO Unmanned logistics — resupply under fire, contested CASEVAC support

★ HQ Command node — protected C2, priority target for adversary decapitation

AI PIPELINE

Four-Stage c-UxS Analysis Pipeline

Each run executes a deterministic sequence. No single point of failure. Each stage produces a structured JSON artifact consumed by the next — fully auditable at every step.

Force Digitization

Parse blue/red unit disposition. Classify by type, side, position. Build structured force object. Deterministic — no model inference required.

<1s

Threat Assessment

AI evaluates adversary force composition. Identifies priority targets, threat level (HIGH / MED / LOW), blue force vulnerabilities, and terrain factors. Returns structured threat object.

15–25s

COA Generation

AI generates three distinct courses of action (alpha / bravo / charlie) — each with concept, assigned tasks per unit type, risk assessment, and tactical advantage. Output: structured COA array.

25–40s

Adversarial Scoring

Red-team AI scores each COA on feasibility, lethality, risk, and mission success. Identifies winner and provides rationale. Adversarial framing eliminates confirmation bias.

15–25s

Offline-first by design. The entire pipeline runs on edge hardware — no cloud dependency, no data exfiltration. Fine-tuned 3B model runs on phone-class hardware (Snapdragon 8 Gen 3 / Apple A17) with no internet connectivity required. All inference local.

VALIDATED SCENARIO PROFILES

Three Core c-UxS Scenario Types

Validation runs covered a range of operational contexts, terrain conditions, and force compositions. Each scenario type tests distinct threat dynamics.

SCENARIO 01

Urban c-UAS

FPV and UAS threat against blue HQ node in built environment. Constrained maneuver space. High RF density. EW and C_SUAS primary response.

FPV UAS C_SUAS EW SHORAD

SCENARIO 02

Open Desert Swarm

Multi-vector SWARM and UAS saturation attack in open terrain. Blue force must allocate DE, SHORAD, and EW against distributed threat without massing.

SWARM UAS DE SHORAD RECON

SCENARIO 03

Littoral UxS

USV and UAS combined threat in coastal/riverine zone. Interoperability between naval and ground c-UxS assets. Coordination of EW with kinetic response.

USV UAS EW C_SUAS HQ

ENDURANCE VALIDATION

24-Hour Continuous Pipeline Test

The c-UxS pipeline was subjected to a 24-hour continuous endurance run to validate stability, consistency, and security gain across extended operations — simulating sustained contested-environment decision support.

ENDURANCE TEST — PASS — 24H CONTINUOUS RUN

Automated execution with monitored outputs. Failure defined as pipeline crash, invalid COA structure, or security gain ≤ 0.

17,263 Total runs

0 Failures

+0.296 Mean security gain

100% Positive gain rate

7.6s Median latency

98% P1 advantage rate

FINDING — ZERO PIPELINE FAILURE OVER 17,000+ RUNS

No crash, hang, invalid output, or negative security gain was observed across the full 24-hour run. The pipeline degraded gracefully on edge cases (incomplete force data, sparse scenarios) without producing invalid COA structures.

FINDING — 92.4% INTENT ACCURACY (n=3,000, 5-FOLD CV)

Commander intent classification measured at 92.4% ±10 using 5-fold cross-validation on 3,000 labeled c-UxS scenarios. Metric reflects the system's ability to correctly identify the commander's operational intent from the scenario context — the foundation of doctrinal COA alignment.

PERFORMANCE PROFILE

c-UxS vs Conventional Domains

The c-UxS engine shares the same adversarial pipeline architecture as land, naval, and air domains but is tuned for the higher unit count, shorter engagement timelines, and lower cost-per-unit economics of unmanned systems operations.

Characteristic	Land / Naval / Air	c-UxS Domain
Typical unit count per engagement	4–12	10–50+
Engagement timeline	Hours to days	Minutes to hours
Decision cycle requirement	Standard (JP 5-0 MDMP)	Compressed (RDSP)
COA pipeline latency	~79s (Rapid mode)	~55–90s
Model requirement	3B fine-tuned	1.5B–3B (edge capable)
Validated scenarios	300 (canonical set)	17,263+
Hardware floor	Tactical tablet	Phone-class (1.5B GGUF)

DEPLOYMENT

From Server Rack to Smartphone

The c-UxS module is the first ConclAive domain designed to run natively on phone-class hardware — addressing the forward-edge deployment requirement where server infrastructure is unavailable. A 1.5B quantized model (Q4_K_M GGUF) delivers full three-stage pipeline inference on a consumer smartphone with no connectivity requirement.

Deployment Tier	Hardware	Model	Pipeline Time	Notes
Edge Server	RTX 3090 / A100	14B fine-tuned	~15–30s	Full pipeline, MLX GPU
Tactical Tablet	M-series Apple Silicon	3B fine-tuned	~55–80s	Offline, air-gapped
Smartphone	Snapdragon 8 Gen 3	1.5B GGUF (Q4)	~60–90s	Forward edge, no infra

llama.cpp inference stack. Phone deployment uses llama.cpp via Termux on Android, serving OpenAI-compatible API locally at port 11434. Static tile maps (SAT/TOPO/OSM). No internet dependency at any layer. Full COA pipeline runs end-to-end on device.

Request c-UxS Technical Brief

Architecture deep-dive, fine-tuning methodology, deployment guide, and live demonstration available on request to qualified defense and government stakeholders.

CONTACT — INFO@LXILC.COM