Maasdorp, Francois De Villiers. Doppler-only target tracking for a multistatic radar exploiting FM band illuminators of opportunity. PhD Thesis. Department of Electrical Engineering, University of Cape Town, 2015.
Commensal Radar (CR), defined as a subclass of Passive Radar (PR), is a receive only radar that exploits non-cooperative illuminators of opportunity for target detection, location and subsequent tracking.
The objective of this thesis is to evaluate the feasibility of using a Frequency Modulation (FM) Broadcast band CR system as a cost effective solution for Air Traffic Control (ATC). An inherent complication by exploiting FM is the low range resolution due to the low bandwidth of FM radio signals. However, due to typical long integration times associated with CR, the frequency domain resolution is typically very good. As a result, measurements of the target’s Doppler shift are highly accurate and could potentially make FM illuminators a viable source for ATC purposes. Accordingly, this thesis aims to obtain a comprehensive understanding of using high resolution Doppler measurements to accurately track the position of a target.
This objective has been addressed by performing a comprehensive mathematical analysis for a Doppler only tracking CR system. The analysis is verified with a tracking simulation, in which the Recursive Gauss Newton Filter (RGNF) is used and lastly, a field experiment was conducted to produce tracking results based on real measurement data.
Results demonstrated that Doppler only target tracking from real measurement data is possible, even when the initial target state vector is initialised from real measurement data. A good degree of correlation is achieved between the theoretical, simulated and measured results, hence verifying the theoretical findings of this thesis. Ensuring that the observation matrix is properly conditioned in Doppler only tracking applications is important, as failure to do so results in tracking instability. Factors that influence the conditioning of the observation matrix are; the number of receivers used (assuming the basic observation criteria is met) and the placement of the receivers, keeping in mind the possibility of Doppler correlation in the measurements.
The possibility of improving an ill-conditioned observation matrix is also demonstrated. In general, tracking filters, for example the RGNF, typically employ time history information and therefore, a direct comparison to the Cramer Rao Lower Bound (CRLB) is unrealistic and accordingly a new theoretical lower bound, called the Cumulative CRLB was derived that does account for time history measurements.
Although the best results for this thesis are achieved by using long integration periods (4 s), the effect of Doppler walk was not compensated for and is an aspect that requires further investigation to potentially further improve on the results obtained in this thesis.
As a final conclusion for this thesis; the Doppler only target tracking delivered some encouraging results, however a qualification test in the form of an extensive trial period is next required to motivate Doppler only tracking for ATC purposes.