HARAS · UNIT B · AI SENSOR-FUSION & C2 APPLIANCE

The mind
we are building.

A planned edge-AI appliance being engineered to fuse every RASAD track, every third-party radar feed, and every RF signature into one clear picture for the operator — and cue the right effector, in the right sector, at the right moment.

Sensor inputs

RASAD masts (N) 3rd-party radars RF DF network Camera feeds

→

HARAS

→

Effector cues

SAMT jammer array SAQR interceptor DIRAA (partner) Operator authority

§ 01— Four Fusion Pillars

Four ways HARAS is being built
to turn noise into signal.

— MULTI-SENSOR FUSION

N sensors, one track.

Design goal: probabilistic Kalman/JPDA fusion across RF, radar, EO/IR and acoustic — assigning each detection a confidence, and each confirmed track a single identity across sensors.

TARGET · < 250 ms end-to-end

— THREAT CLASSIFICATION

Drone, not a bird.

An in-house AI model, trained on regional signal libraries, is being designed to classify targets — drone / bird / helicopter / balloon — with a false-positive rate low enough for operator confidence.

TARGET · < 2 % FALSE POSITIVE

— EFFECTOR CUEING

Right tool. Right sector.

Design goal: real-time route each confirmed threat to the appropriate effector — SAMT for soft-kill, SAQR for kinetic, DIRAA partner for last-resort — with human-in-the-loop authority.

TARGET · SUB-SECOND HAND-OFF

— OPERATOR UI

One glass. One decision.

A planned single-pane operator console — Arabic and English, dark-optimised, designed for 30-minute-cognitive-load shifts in a real Gulf ops room. No 27 windows. No modal fatigue.

TARGET · 1 OPERATOR : N SITES

§ 02— Key Specifications

Engineered for
a sovereign ops floor.

Every value below is a design target. No standards conformance or benchmark has yet been obtained.

Form factor	2U rackmount TARGET · 19″ RACK
Compute	ARM + GPU accelerator CANDIDATE
Fusion capacity	Up to 32 sensors TARGET · PER APPLIANCE
Concurrent tracks	≥ 500 TARGET
Ingest interfaces	SFP+ · Cat-6A · MQTT TARGET
Open C-UAS API	REST + gRPC TARGET · V1
Cyber posture	UAE NESA / IEC 62443-4-2 TARGET · NOT YET CERTIFIED
Data sovereignty	On-prem · air-gap capable DESIGN INTENT
Redundancy	Dual PSU · active-standby pair TARGET

Why HARAS is designed
to sit at the centre.

— NO VENDOR LOCK-IN

Open C-UAS API being designed to accept any compliant sensor and cue any compliant effector — third-party or RADI.

— SOVEREIGN BY DEFAULT

Every model, every track, every log intended to stay on-prem. Air-gap deployment is a design goal for MoD-tier installations.

— DESIGNED FOR THE GULF

Model libraries being trained on Gulf-region RF and radar signatures — not on borrowed European datasets that miss the drones we actually see.

§ 03— Integration

Speaks every dialect.
Answers to yours.

HARAS is being designed to sit at the centre of any counter-drone architecture — sovereign, coalition, or enterprise. The planned Open C-UAS API is intended to accept telemetry from any compliant sensor, and to cue any compliant effector, over mTLS-encrypted channels. Integration with UAE MoD IADS, Tawazun-programme partners, and third-party C2 platforms is being scoped ahead of first evaluation.

The mindwe are building.

Four ways HARAS is being builtto turn noise into signal.

N sensors, one track.

Drone, not a bird.

Right tool. Right sector.

One glass. One decision.

Engineered fora sovereign ops floor.

Why HARAS is designedto sit at the centre.

Speaks every dialect.Answers to yours.

HARAS is the brain.Every RADI product feeds it.

The mind
we are building.

Four ways HARAS is being built
to turn noise into signal.

Engineered for
a sovereign ops floor.

Why HARAS is designed
to sit at the centre.

Speaks every dialect.
Answers to yours.

HARAS is the brain.
Every RADI product feeds it.