SUN ETD - Theses and Dissertations

Permanent URI for this community

https://scholar.sun.ac.za/handle/10019.1/804

This community is a clearing house for masters and doctorates submitted via htp://etd.sun.ac.za

Browse

Browsing SUN ETD - Theses and Dissertations by Author "Van Zyl, Christiaan"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    A software defined radio based radar system with applications in the detection of wildlife poaching
    (Stellenbosch : Stellenbosch University, 2023-12) Van Zyl, Christiaan; Niesler, Thomas; Stellenbosch University. Faculty of Engineering. Dept. of Electrical and Electronic Engineering.
    ENGLISH ABSTRACT: This thesis presents the development and testing of a low-cost and low-power software-defined r adio ( SDR) b ased r adar p latform f or u se i n anti-poaching applications. A low frequency of 1.1 GHz is specifically selected to investigate the viability of classifying human activity, whilst enabling foliage penetration. A software interface and control tool for the SDR was developed to enable the rapid development and testing of radar techniques, and the collecting and storing of data in the form of I/Q samples. A signal processing pipeline was implemented using the Julia programming language to generate range-Doppler maps from these I/Q samples. Existing non-linear frequency modulated pulse compression waveforms did not yield satisfactory performance due to the low time-bandwidth product regime in which the radar system operates. A pulse compression waveform based on optimised Bézier curves was therefore developed and shown to offer improved performance over existing methods that require a higher time bandwidth product. The best Bézier vertices were determined by a particle swarm optimisation method, which is also presented. A dataset was collected to evaluate the system’s ability to classify targets based on micro-Doppler frequencies. A novel convolutional neural network configuration t hat u ses t emporal D oppler f rames i s p resented a nd s hown to achieve a classification accuracy of 91.6% when testing on a dataset containing a single person. Therefore, it may be concluded that SDR-based radar merits further consideration as a low-cost and low-power approach to the detection of poachers in automated anti-poaching systems.