A passive micro-Doppler sensor for fused situational awareness

Context

An active radar announces its own position. For a sensor intended to sit forward, or to remain useful once an area is contested, that is not a detail of the specification — it is the specification. A passive approach removes the emission and, with it, the straightforward means of measuring range, so everything the sensor knows has to be recovered from signals it did not originate.

Micro-Doppler is what makes the trade worthwhile. Rotating and reciprocating parts — rotor blades, propellers, wingbeat — modulate a return in ways that are characteristic of the thing producing them, which is how a multirotor is distinguished from a bird at a distance where neither is more than a few pixels to anything optical.

Approach

We took the prototype from schematic to working system: board design, component selection, firmware, and the fusion software above it. Component selection was a larger part of the engagement than it would ordinarily be. In this domain, availability, lead time and provenance are design constraints with the same standing as performance, and a part that is superior on its datasheet but single-sourced through an unacceptable route is not a candidate.

The classifier’s output was designed for fusion from the start. A sensor of this kind is rarely the only one present, and its value is in contributing what the others cannot — a signature-based classification with an honest confidence attached, rather than a track asserted with more certainty than the physics supports. Fusing a confident wrong answer degrades the shared picture more than contributing nothing.

Outcome

A working prototype: hardware, firmware and fusion software, classifying targets from micro-Doppler signatures without transmitting. The signal chain is documented to the level that the client’s team can extend the signature library themselves, which is where the ongoing work in passive classification sits.