Thread by @sambendett

1/ QUICK TAKE by Rus mil bloggers who translated a paper on identifying and tracking fiber-optic drones in flight: "FPV drones utilizing fiber-optic links possess zero electronic signature, rendering them invisible to conventional electronic intelligence (ELINT) systems." t.me/VBiblioteka/17…

2/ "However, a drone remains a physical object. It can be detected using passive radar techniques based on SDR receivers—which leverage ambient radio illumination from DVB-T, GSM, or LTE towers—as well as through the use of specialized short-range radars."

3/ "Under these conditions, the key tool for ID-ing is the analysis of micro-Doppler signatures. Traditional radar systems are unable to effectively distinguish small UAVs from birds due to their similar radar cross-sections (RCS)."

4/ "Micro-Doppler analysis solves this problem by analyzing the kinematics of minute movements. How it works and what it offers:"

5/ "Precise Discrimination: The reflected radio signal is modulated by the target's moving parts. The high-frequency rotation of a quadcopter's propellers generates a unique pseudo-Doppler shift pattern, which—when visualized on a spectrogram—differs radically from the low-frequency flapping of bird wings or the swaying of foliage."

6/ "Hover Detection: This technology enables the detection of hovering or autonomously operating aircraft. Even if the radial velocity of the airframe itself is zero, the rotating rotors continue to generate a distinct radio signature."

7/ "Threat Analysis: The system extracts data regarding rotor rotation frequency and dimensions, aiding in the classification of the UAV type, the tracking of drone swarms, and the generation of precise targeting data for engagement systems."

8/ "AI/ML Integration for Automation: The effectiveness of this method increases exponentially when machine learning is implemented (see our previous publications regarding TorchSig."

9/ "The Short-Time Fourier Transform (STFT) converts radio signals into time-frequency spectrograms, which are then processed by convolutional neural networks in a manner analogous to computer vision tasks."

10/ "In real time, AI compares the resulting patterns against an extensive database of signatures, thereby enabling the automatic classification of objects. This critically reduces the rate of false positives and alleviates the cognitive burden on air defense operators."

11/ "A fiber-optic drone—otherwise completely immune to electronic warfare—inevitably reveals its presence in the radio spectrum through the very physics of its flight; however, the detection range remains constrained by the power output of the emitter and the computational capabilities available at the receiver end."