Radio Frequency Fingerprinting

Radio-based navigation aids are widely used in (civil) aviation to enable reliable and safe navigation.  For example, Global Navigation Satellite Systems (GNSS) like GPS or GALILEO are used for geopositioning, but there are also aids to support airport operations using differential GNSS or using Instrument Landing Systems (ILS) to provide the glide path during landing in inclement weather.  These systems utilise different radio frequencies (RF) within the electro-magentic spectrum to provide these services to aircraft.

Due to the increased use of wireless technologies, however, the common frequency spectrum, in which these radio-based navigation systems operate, becomes increasingly crowded and the risk of interference with other transmitters and their signals in the same or close RF bands increases, which in turn might render the navigation aids inoperable.

Therefore it is necessary to detect and identify transmitters that are (or might be) interfering. This can be done using RF fingerprinting, which allows transmitter identification by learning unique hardware-based characteristics of the transmitters e.g. frequency, amplitude, phase, pulse width etc., through the in-phase (I) and quadrature components (Q) of the signals emitted by the interfering transmitters.

The goal of this project is to analyse and cluster GBs of datasets to identify, “fingerprint”, radio transmitters. This is illustrated in the example for vision based methods using spectrograms, but the methods to be used in the project are up-to you. Any classical algorithm, AI or a combination of both is welcome. A benchmarking of different methods is highly appreciated.