MEng Dissertation: Stephen Middleton

 

Profile:

Stephen Middleton
Click here to read the interview

Stephen Middleton grew up in Johannesburg and matriculated from St Benedict’s College in Johannesburg.

After obtaining a BSc in Electrical Engineering from the University of Cape Town, he completed the MEng, specialising in Radar and Electronic Defence. This included attending various radar orientated courses, including clutter modelling and analysis, target tracking, radar signal processing and radar imaging.

The topic of his thesis, which was supervised by Prof Mike Inggs, was “Target tracking in the range-Doppler space”.

He is currently based in Cape Town and working at Von Seidels, an intellectual property law firm in Century City, Cape Town.

Click on the photo to read an interview with him, which is part of the ‘Meet our Alumni‘ series.

 

Citation:

Middleton, Stephen. F  Target tracking in the Range-Doppler space. MEng (specialising in Radar and Electronic Defence) Dissertation. Department of Electrical Engineering, University of Cape Town, 2012.

 

Abstract:

Adobe-PDF-download
Download PDF of Dissertation

Commensal radars make use of transmitters of opportunity to detect, locate and track targets in a manner that is non-detrimental to these transmitters. The target detections are displayed on amplitude-range-Doppler plots and follow curved trajectories with time. In the case of FM band commensal radars, these detections are subject to low and fluctuating range resolution making them difficult to follow visually. This complication can be alleviated by tracking the targets in the range-Doppler space.

This project compares the Kalman, polynomial and recursive Gauss-Newton tracking filters for this purpose by using simulated and real data. The filter performance is evaluated on: tracking errors, computational load, data association statistics and real data tracking. All three filters perform well, however the recursive Gauss-Newton filter tracks the most targets in the real data, achieves low errors and is the most efficient in terms of computational load.

 

 

Leave a Reply