Automatic signals intelligence analysis


This project explores how data from radio frequency sensors can be transformed effectively into useful intelligence about the environment.


  1. The research will develop low complexity techniques to convert radio signals into useful metadata for further processing in a sensor network.
  2. Create new approaches to modulation and protocol identification with a particular focus on a favourable performance vs complexity trade-off.
  3. Undertake metadata analysis to distinguish normal operation from anomalous events, such as the presence of a signal at an unexpected time or frequency.


Wideband radio signals intelligence systems generate large volumes of received data, as well as the metadata associated with the constituent narrowband signals. Thousands of narrowband signals are simultaneously detectable at each site in the sensor network. Given that these signals can be detected, intercepted and processed, the problem becomes: What should be done with the data to maximize the intelligence garnered from it? This project will address this question as follows: firstly, the underlying ‘normal’ behaviour of the electromagnetic environment needs to be identified based on the collected waveforms. This will involve generating metadata that can help to delineate between normal and anomalous traffic events. Secondly, fast classification techniques should be developed based on the radio signal properties to identify signal modulation or protocol formats in use. Thirdly, automated analysis techniques should be developed for placing signals into different priority classes for further investigation by a skilled human operator.

Research theme: 

Industrial partner: 


Principal supervisor: 

Professor James Hopgood
University of Edinburgh, School of Engineering