How to Use a Drone-Mounted BLE Scanner for Search and Rescue (DIY and Off-the-Shelf Options)

ByPriya Nair

Yes, This Works — and There Are Already Products for It

The concept is solid, academic research backs it, and there are commercial payloads built for exactly this use case. The main decision is whether to buy one or build one.

Commercial Options Built for SAR Signal Detection

Parsons, through their QRC Technologies subsidiary, makes a product called BlueFly that is purpose-built for drone-based SAR detection. It is a drone-agnostic payload — compatible with nearly any commercial or government UAV — and scans for Bluetooth Low Energy and Wi-Fi signals simultaneously. Detection range reaches up to 200 meters, and the unit weighs under a pound, so payload impact is minimal.

What separates it from a basic BLE scanner is the range of devices it catches. Beyond smartphones, it detects smartwatches, fitness trackers, wireless earbuds, and Bluetooth-enabled medical devices like glucose monitors and pacemakers. That last category is significant: if someone is unconscious or incapacitated, a pacemaker or glucose monitor may be the only signal they’re emitting. Detection data streams in real time to Team Awareness Kit (TAK)-enabled Android devices, so every team member sees signal hits on the same shared map as the drone passes overhead.

X-SAR, from X-Sensor, is another modular system available in both handheld and drone-mountable configurations with integrated Wi-Fi and Bluetooth detection. It runs on swappable external power banks for extended field deployments.

Neither product lists public pricing. Both are aimed at agency buyers — expect a quote process rather than an online checkout.

What the Research Actually Shows

Researchers have been testing this for several years. A peer-reviewed paper in Engineering, Technology and Applied Science Research demonstrated that a BLE-equipped drone could scan a 3,600 square meter area in under five minutes using a Raspberry Pi as the receiving hub. A separate project called SARDO combined pseudo-trilateration with machine learning to locate a phone to within a few tens of meters — no cellular network needed, no app on the victim’s device.

The reason it works is straightforward. Modern smartphones broadcast BLE advertisement packets nearly constantly to support features like AirDrop, Nearby Share, and Find My. A drone flying above a tree canopy has far better line-of-sight to those signals than a ground team pushing through brush. Each pass gives you signal strength at a known GPS coordinate; multiple passes give you a heat map.

Building It Yourself with a Raspberry Pi

If a commercial payload is outside your budget or procurement timeline, a DIY build is genuinely within reach. A Raspberry Pi 4 or 5 with its onboard Bluetooth hardware — or a USB BLE dongle for better sensitivity — is enough to start. The Python library bleak (available on PyPI) handles asynchronous BLE scanning on Linux and returns MAC addresses, RSSI values, and raw advertisement payloads for every detected device.

A minimal scan loop looks like this:

import asyncio
from bleak import BleakScanner

async def scan():
    devices = await BleakScanner.discover(timeout=5.0)
    for d in devices:
        print(d.address, d.name, d.rssi)

asyncio.run(scan())

Pair that output with GPS timestamps from the drone’s telemetry feed and you have geo-tagged signal logs. Run a grid survey over a defined area and you can build a signal density heat map. Dense clusters indicate where to send the ground team first.

Antenna Selection

The Pi’s built-in chip antenna is fine for a workbench but limited for aerial work. For downward-facing scanning during a grid pass, a 2.4 GHz patch antenna mounted face-down improves gain meaningfully over an omnidirectional whip. BLE operates in the 2.4 GHz band, identical to Wi-Fi, so standard 2.4 GHz antennas work directly without adapters.

For a direction-finding follow-up pass once you’ve narrowed the search area, a Yagi-style directional antenna on a horizontal gimbal can help close in on the source. That’s a more complex rig, but it’s the step that takes you from “probably in this quadrant” to “within 20 meters of that tree line.”

Flight Mission Planning

Mapping software like DJI Ground Station Pro, Litchi, or UgCS supports automated grid survey missions. You define the area boundary, set altitude and overlap, and the drone flies the pattern autonomously. That gives you consistent altitude, systematic coverage, and repeatable passes — exactly what you need to build a reliable heat map.

Altitude is a direct tradeoff between coverage and signal strength. Higher means more ground per pass but weaker received signal. For BLE, 30 to 60 meters above ground is a reasonable starting range for open terrain. Under a forest canopy you may need to fly lower, which means more passes to cover the same area. Plan mission altitude based on terrain type, not a single fixed number.

The Accuracy Problem with RSSI

RSSI (Received Signal Strength Indicator) is noisy. BLE signals reflect off rocks, absorb into wet soil, scatter through dense vegetation, and attenuate differently depending on whether the human body is between the device and the antenna. One reading tells you “a device is somewhere nearby” — not a coordinate.

Outdoor environments are much harder than the structured indoor settings where BLE localization performs best. Realistic expectations for a field deployment: narrowing a large search area — say several hundred acres — down to a zone small enough for ground teams to cover efficiently. That’s a meaningful operational win even without sub-meter accuracy.

Averaging many readings across multiple passes reduces noise considerably. Flying slower and logging more samples per unit of ground area helps too. More sophisticated builds apply Kalman filtering or train a model on local environmental signal variation. That’s beyond a first build, but the option exists if accuracy requirements grow.

What Devices You Will and Won’t Detect

Reliably detectable: iPhones and Android phones (both advertise BLE nearly continuously), Apple AirTags, Tile and Samsung SmartTag trackers, most smartwatches and fitness bands, wireless earbuds, and Bluetooth medical devices.

Less reliable: standard car key fobs (they use dedicated RF protocols, not BLE), older IoT sensors that only advertise intermittently, and any device with Bluetooth manually disabled.

One behavior worth understanding up front: both iOS and Android rotate their BLE MAC address roughly every 15 minutes for privacy. The same phone appears under a different address on each scan pass. For SAR this doesn’t matter — you’re detecting any device in the area, not tracking a specific one by ID — but it does mean your log analysis needs to avoid double-counting the same device across passes when estimating the number of people present.

Sources


Similar Posts