Unmanned Aerial Vehicle Intrusion Detection

Abstract

A system detects and identifies unmanned vehicles (UVs) from radio communications between UVs and their controllers. One or more radio frequency (RF) signal detectors can detect RF signals, including downlink signals transmitted by a UV or uplink signals transmitted by a UV controller. A feature extractor can extract signal features from detected RF signals, and a classifier performs machine learning to identify at least one of the UV and the UV controller based on the signal features. Machine learning can employ an artificial neural network, which may perform deep learning.

1 . A system for unmanned vehicle (UV) detection in a radio frequency (RF) environment, comprising:
 one or more radio frequency (RF) signal detectors configured to detect RF signals, including RF signals transmitted by at least one of a UV and a UV controller;   a feature extractor configured to extract signal features from detected RF signals; and   a classifier that employs machine learning to identify at least one of the UV and the UV controller based on the signal features.   
     
     
         2 . The system of  claim 1 , further comprising a filter coupled between the one or more RF signal detectors and the feature extractor, the filter configured to remove one or more unwanted frequencies from the detected RF signals, temporally filter the detected RF signals to retain only temporally relevant signals, or spatially filter the detected RF signals. 
     
     
         3 . The system of  claim 2 , further comprising a computer-readable memory for storing filtered RF signal data for later analysis. 
     
     
         4 . The system of  claim 1 , further comprising a decoder coupled between the one or more RF signal detectors and the feature extractor, the decoder configured to decode at least one of the detected RF signals. 
     
     
         5 . The system of  claim 4 , wherein the decoder is configured to perform blind-adaptive decoding. 
     
     
         6 . The system of  claim 1 , wherein the classifier is configured to compare one or more of the signal features to known signal features in order to distinguish a signal of interest from other signals. 
     
     
         7 . The system of  claim 1 , wherein the classifier is configured to receive a device identifier from the detected RF signals, and comparing the device identifier to an authentication database to determine if the at least one of the UV and the UV controller is permitted. 
     
     
         8 . The system of  claim 1 , wherein the one or more RF signal detectors are responsive to a sensor output indicating an incursion of a perimeter or area by a UV for initiating signal detection. 
     
     
         9 . The system of  claim 1 , wherein the classifier comprises an artificial neural network. 
     
     
         10 . The system of  claim 1 , wherein machine learning comprises deep learning. 
     
     
         11 . An apparatus, comprising:
 a radio receiver coupled to at least one antenna and configured to detect radio frequency (RF) signals communicated between an unmanned vehicle (UV) and a UV controller;   a memory; and   one or more processors operatively coupled to the memory and the radio transceiver, the one or more processors configured to:   extract signal features from detected RF signals; and   employs machine learning to identify at least one of the UV and the UV controller based on the signal features.   
     
     
         12 . The apparatus of  claim 11 , wherein the one or more processors are configured to remove one or more unwanted frequencies from the detected RF signals, temporally filter the detected RF signals to retain only temporally relevant signals, or spatially filter the detected RF signals. 
     
     
         13 . The apparatus of  claim 12 , configured for storing filtered RF signal data in the memory for later analysis. 
     
     
         14 . The apparatus of  claim 11 , wherein the one or more processors are configured to decode at least one of the detected RF signals. 
     
     
         15 . The apparatus of  claim 14 , wherein the one or more processors are configured to perform blind-adaptive decoding. 
     
     
         16 . The apparatus of  claim 11 , wherein the one or more processors are configured to compare one or more of the signal features to known signal features in order to distinguish a signal of interest from other signals. 
     
     
         17 . The apparatus of  claim 11 , wherein the one or more processors are configured to receive a device identifier from the detected RF signals, and comparing the device identifier to an authentication database to determine if the at least one of the UV and the UV controller is permitted. 
     
     
         18 . The apparatus of  claim 11 , wherein the one or more processors are responsive to a sensor output indicating an incursion of a perimeter or area by a UV for initiating signal detection. 
     
     
         19 . The apparatus of  claim 11 , wherein the one or more processors comprises an artificial neural network. 
     
     
         20 . The apparatus of  claim 11 , wherein machine learning comprises deep learning.