EchoCore AI · Sentinel Shield · IDEaS POINT 2026
An elevated multi-modal sensing architecture designed to detect silent, emission-free UAVs that evade ground-based radar. Built for the DND Sentinel Shield challenge.
The Problem
"Detection of Class I and II UAS, including designs that do not rely on GPS and Radio-Frequency." DND Sentinel Shield Challenge · IDEaS POINT · 2026
RF Monitoring Fails
Silent UAVs emit nothing. Fiber-optic controlled drones are completely invisible to RF sensors.
Ground Radar Blinded
Low-angle geometry means buildings, vehicles and terrain clutter mask small targets with RCS as low as 0.01 m².
The MTI Problem
Ground-based Moving Target Indication fights side-lobe clutter at shallow angles. There is no clean solution at ground level.
The Requirement
10–30 km radius · 24/7 continuous · RF-silent capable · Class I & II UAS · −40°C to +40°C
Our Approach
"From above, ground clutter is static. Only moving objects break the Doppler silence. The target reveals itself."
Ground clutter is nearly stationary. From elevation, static terrain, buildings, and parked vehicles sit at near-zero Doppler — Moving Target Indication (MTI) filters them out. A moving UAV breaks that Doppler silence and stands out, regardless of RF emissions. This is the same physical principle that has made airborne downward-looking radar operationally effective for over 40 years.
Ground-based MTI fights shallow-angle clutter and terrain masking. From elevation, static ground returns concentrate near zero Doppler, and moving aerial targets stand out against a more predictable background. Micro-Doppler then separates rotor harmonics from bird wingbeats. The problem is not solved — it is made more tractable.
Tethered aerostat radar is not a novel idea. The U.S. Tethered Aerostat Radar System (TARS) has operated continuously since 1978 — eight sites along the U.S.-Mexico border, each carrying a downward-looking radar at up to 15,000 ft AGL, detecting low-altitude targets that ground radar misses due to terrain masking. (Source: Wikipedia TARS / CBP Official)
QinetiQ's current commercial aerostat platform explicitly supports Class I/II/III UAV detection with radar payloads at 15,000 ft. (Source: QinetiQ Aerostat Solutions)
EchoCore is building a modern, smaller-scale, UAS-optimized implementation of a concept with four decades of operational proof.
Tethered: infinite endurance, ground power, 24/7 persistent coverage. Untethered: active patrol, extending coverage from 10 km to 30 km+ radius as the platform repositions.
OPERATIONAL ARCHITECTURE · Subject to flight validation
Compared to fixed-wing ISR aircraft or satellite surveillance, the platform cost is a fraction. Helium-sustained lift eliminates fuel burn and dramatically reduces operating costs compared to fixed-wing ISR platforms. Tethered power eliminates battery limits. A single platform can maintain station for weeks, not hours.
Commercial FPV drones and threat UAVs typically operate below 500m AGL. The surveillance platform operates at 500–3000m, above the typical threat envelope, with unobstructed sightlines across the full surveillance zone.
The target architecture integrates three sensing layers. Each is designed to fill the gaps the others cannot.
Primary detection. Detects ALL moving objects regardless of RF emissions. MTI filtering removes static clutter. Micro-Doppler distinguishes rotor harmonics from bird wing-flap.
Range 10–30 km · 24/7 all-weather · RF-silent capable
When a drone transmits, RF provides instant identity. Complements radar without depending on it.
Passive · No emissions · 2.4 / 5.8 GHz · Remote ID compliant
High-confidence identity confirmation once radar has acquired and tracked target. Day and night capable.
Identity confirmation · Day/night · Form factor recognition
The architecture routes all three sensor streams into a shared AI pipeline. Confidence-weighted fusion produces target class, confidence score, and track state. CoT/SAPIENT output format is the design target for C2 integration.
Multi-hypothesis reasoning · Confidence scoring · C2 output
AI Classification Pipeline
The pipeline follows established radar signal processing practice. Each stage maps to well-understood DSP and ML techniques, with no black boxes and no hand-waving.
Raw I/Q Samples
Signal acquisition · ADC 100+ Msps · complex baseband
Range Processing
Pulse compression · CFAR · MTI filter
Time-Freq Transforms
Range-Doppler map · 2D FFT · ~0.5 m/s bin
Micro-Doppler Spectrogram
STFT · blade harmonics · JEM signatures · 10–80 ms windows
Feature Extraction
RCS statistics · Doppler harmonics · spectral entropy
Classifier Ensemble
SVM · CNN · Platt scaling · confidence calibration
Decision Fusion
Bayesian · Dempster-Shafer · multi-hypothesis tracking
C2 Output
CoT / SAPIENT / ATAK · geo-referenced tracks
Classification accuracy on micro-Doppler spectrograms: >95% UAV vs. bird (published benchmark, side-looking geometry, X-band, DJI Phantom class targets). Top-down geometry validation in progress. Pipeline architecture follows DRDC CUAS signal processing doctrine.
Intellectual Property
Development Roadmap
A staged validation approach aligned with standard aerospace TRL methodology and with the Sentinel Shield procurement timeline. Each phase builds on demonstrated capability. No speculative jumps.
Sentinel Shield · IDEaS POINT Official Schedule
Q2–Q3 2026 · Active now
Sep 2026 – Apr 2027 · Sentinel Shield Build phase
May 2027 test · FY 27/28 delivery · 27 CAF sites
"The platform concept has 40 years of operational precedent. The sensors exist. The algorithms are being built. This is a modern, UAS-optimized integration — not a speculative architecture."
EchoCore AI · CA Patent Application 3,286,560 · CUAS 2025 & 2026 participant · IDEaS Sentinel Shield applicant · June 24, 2026 proposal deadline
EchoCore AI is developing this system for the DND Sentinel Shield challenge (IDEaS POINT, deadline June 24, 2026). We are looking for partners with expertise in radar signal processing, AI/ML, and aerial systems.