Perimeter Defense & Drone-Based Surveillance

Critical infrastructure perimeters present unique detection challenges. Facilities may span thousands of meters of fence line through varied terrain—open fields, wooded areas, waterfronts, and adjacent public roadways. Traditional perimeter security relies on CCTV and periodic guard patrols, creating coverage gaps both spatial (blind spots between cameras) and temporal (patrol intervals of 30-60 minutes). At a 2-kilometer perimeter, a determined intruder can breach, transit the detection zone, and reach critical assets within 3-5 minutes—far less than the typical patrol interval.

Single-technology perimeter detection systems suffer from unacceptable false-alarm rates in outdoor environments. Fence-mounted vibration sensors trigger on wind, wildlife, and vegetation contact. Outdoor motion cameras alert on shadows, headlights, and thermal shimmer. Buried cable sensors activate during ground frost heave and heavy vehicle traffic on adjacent roads. Each false alarm requires operator attention, guard dispatch, or both—and the resulting alarm fatigue is the primary reason genuine perimeter breaches are missed.

Regulatory frameworks for critical infrastructure demand specific perimeter security performance levels. NERC CIP-006 for bulk electric system facilities requires defined Electronic Security Perimeters and Physical Security Perimeters with monitored access points. CFATS for chemical facilities mandates risk-based security performance standards including detection, delay, and response. NRC 10 CFR 73 for nuclear facilities specifies detection probability thresholds and assessment timelines. Any perimeter defense system must demonstrate measurable performance against these standards.

We design layered perimeter defense architectures with three concentric detection zones. The outer zone (50-200 meters beyond the fence line) uses ground-surveillance radar (Magos, Navtech, or Axis D2210-VE) to detect and track moving targets in all weather conditions, classifying them as human, vehicle, animal, or environmental. The mid zone (fence line) uses distributed fiber-optic sensing (DAS) cable bonded to the fence fabric, detecting climbing, cutting, and lifting attempts via acoustic signature analysis. The inner zone (fence line to asset boundary) uses thermal imaging cameras with video analytics for target classification and tracking.

When the outer zone radar detects a human or vehicle target approaching the perimeter, the system cues the nearest PTZ camera to the target's radar-derived coordinates and initiates classification. If the target reaches the fence line and triggers the fiber-optic sensor, the system correlates the radar track with the fence alarm (spatially and temporally) to confirm a breach attempt and elevates the event to a verified alarm. At this point, the system can dispatch an autonomous drone (DJI Dock 2, Skydio X10, or Nightingale Security) to the breach location for real-time aerial visual assessment, providing the operator with live video of the threat before committing a physical response.

The correlation engine fuses data from all sensor layers, maintaining persistent target tracks from detection through assessment and response. Each target is assigned a threat score based on behavior patterns: direct approach to the fence line scores higher than parallel movement, cutting/climbing signatures score higher than leaning/pushing, and post-breach movement toward critical assets triggers maximum priority. The entire detection-classification-assessment pipeline executes within 30 seconds, meeting NERC CIP and CFATS assessment timeline requirements.

The distributed fiber-optic acoustic sensor (DAS) uses coherent optical time-domain reflectometry to detect phase changes in backscattered light along a standard single-mode fiber bonded to the fence fabric. The interrogation unit (OptaSense, Fotech, or Halo) pulses a narrow-linewidth laser into the fiber and analyzes the returning Rayleigh backscatter at 10,000 samples per second per spatial resolution cell (3-5 meters). Classification algorithms running on the interrogation unit's embedded GPU distinguish between climbing signatures (periodic loading with 1-3 Hz cadence), cutting (high-frequency broadband acoustic signature above 2 kHz), lifting (low-frequency sustained loading), and environmental noise (wind: broadband below 500 Hz, rain: stochastic broadband). The DAS system covers up to 40 km of fence line from a single interrogation unit, with fiber routed through weather-sealed conduit along the fence line and terminated in a reflective loop for bidirectional monitoring.

Ground surveillance radar integration uses the Magos ScenarioPlus or Axis D2210-VE radar units, which output target tracks via the ONVIF Profile M analytics metadata interface or vendor-specific REST APIs. Radar tracks include target position (range and bearing, converted to GPS coordinates via known radar mounting position), velocity, radar cross-section (RCS), and vendor-classified target type. Our correlation engine ingests radar tracks and fiber-optic events via MQTT, performing spatial correlation using PostGIS spatial functions: a radar track passing within 10 meters of a fence DAS activation within a 5-second temporal window generates a correlated multi-sensor event with confidence scoring based on the consistency of the spatial and temporal alignment.

Autonomous drone dispatch is orchestrated via the drone manufacturer's SDK (DJI FlightHub 2 API for DJI Dock, Skydio Cloud API for Skydio X10). Upon receipt of a verified perimeter alarm, the system generates a mission flight plan from the nearest drone dock to the alarm location, including waypoints for optimal camera angles considering the breach direction indicated by the radar track. The drone launches within 15 seconds, reaches the breach location within 60-90 seconds (typical for perimeters up to 1.5 km from the dock), and enters an autonomous orbit pattern providing continuous video of the breach zone. Video is streamed via WebRTC to the operator console and simultaneously recorded to the evidence management system. All drone operations comply with FAA Part 107 waiver requirements for BVLOS (Beyond Visual Line of Sight) operations at the facility.