This PhD thesis focuses on quantifying the impact of oscillator phase noise on the design of MMW CW radar systems with the goal of optimising the system to achieve better target detection and tracking. Phase noise in the transmitters of radar systems is known to distort the target response by broadening the linewidth and raising the noise floor of radar systems when a strong scatterer is present in the scene, hence degrading the detection and tracking performance. The situation is worse when multiple large scatterers are present, as the noise sidebands of all scatterers superimpose causing small targets, like pedestrians, to disappear in the phase noise sidebands. Some of the phase noise is cancelled at short ranges in coherent radars but the cancellation is not effective at long ranges.This research presents the design of phase noise reduction techniques. Phase noise modelling at the system level is presented to elaborate the methods of minimising the impact of phase noise. It will be shown that the frequency synthesiser is the most significant phase noise contributor. The design and implementation of a low phase noise signal source is presented. Both linear and non-linear phase noise models are used and developed further in order to meet the radar optimisation goals. An elaborate relationship of the phase spectrum with the RF spectrum of an oscillator is presented. The idea of coherence time is used as a tool for the selection of radar signal sources, and a novel derivation of the minimum bound on the transmitter phase noise level presented to prevent excessive distortion of target spectra.A new phase noise model is developed for the analog-to-digital conversion process using an independent sampling clock. The case of a sampling clock derived from the transmitter's reference oscillator will also be discussed. The models aid the selection of an appropriate sampling clock for a given radar application. A novel method of characterising the phase noise statistics using the integer and the fractional Brownian motion models will be presented. Models for the lineshape and the linewidth of the RF spectrum are dealt with in detail by reviewing the existing models in the literature. These analyses aid in assessing the fundamental resolution capability of radar systems in terms of the phase noise processes. A novel analysis of the RF spectrum of a signal impaired with random-walk phase noise is detailed, and it is shown that the RF spectrum exhibits time-dispersion and satellite peaks. It is shown that the success of the proposed work depends on techniques for careful measurement, analysis, and mitigation of the various noise processes.
|Date of Award
|26 Jun 2017
|Robert Watson (Supervisor) & Stephen Pennock (Supervisor)