SELEX Galileo 04 (2009-13)

Title: Detecting Unknown Waveforms in Time and Space

Research Engineer: Fraser Gordon
Sponsor: SELEX Galileo
Academic Supervision: Prof Bernie Mulgrew, University of Edinburgh

Classically the best way of detecting a particular signal in noise is to use a matched filter. In fact, if the noise is white and Gaussian, it is the optimal procedure. Many communications and radar systems use the matched filter as the first line or corner stone of all detection and decoding.

Electronic warfare (EW) receivers must detect radar pulses and use the measured characteristics of the transmission to identify whether they are associated with potentially hostile systems. An important consideration is that the Electronic Warfare receiver should be able to make this observation before the radar is able to form a track on the host platform – this is known as the range advantage. The disadvantage suffered by the EW receiver is in it not being matched to the transmitted pulse, and the worse the match is the more the range advantage is degraded. Modern EW systems employ digital receiver technology similar to that used in software defined radios. Digital EW receivers offer the ability to improve detection of radar pulses over earlier technology based on video detection, but current techniques have limited efficacy.

The matched filter technique employed by all modern radars is very powerful, inspiring the use of transmitted signals that are observed with degraded signal/noise by the opportunistic listener but which are readily detected by an appropriate listener who has knowledge of the transmitted signal. Faced with an environment of transmitted signals, what can we do as an opportunistic listener who does not know the transmitted waveform and thus cannot access the signal/noise improvements offered by the matched filter?

This is the challenge of the current project. It will be addressed by exploiting multiple realisations of the signal in both time and space and will build on a current Eng.D. project that has considered only single realisations of the signal. Many radar systems transmit multiple pulses, one after another in time. One research question is: given a possible detection of the first pulse can we derive a strategy to improve the probability of detection of the second and subsequent pulses? Many receiving systems employ multiple channels that are spatially separated. This brings us to the second research question: given a possible detection of a signal of interest in one channel of the receiver can we increase the probability of detection in the second and remaining receivers? The problem is multi-faceted and there are many dimensions to the solution space, but all solutions have to be mindful of the constraints of available processor speed, architecture and power. The project will also address the problem of detecting digital communications and wireless network communications.