Abstract
The use of Fast Fourier Transform (FFT) to implement correlation for
fast acquisition has gained its popularity not only in software GPS
receivers but also in some ''hard'' receiver implementations. However,
its use for GPS signal tracking is not well appreciated except for
post-correlation coherent integration and residual Doppler estimation
to assist conventional tracking loops. Common perception fixes on
the idea that the tracking of each GPS satellite in view would require
at least a pair of FFT and IFFT, assuming that the replica FFT can
be precalculated and stored in memory. Such a processing scheme can
indeed incur an expensive computational load in the tracking mode.
In addition, the FFT-calculated correlation only provides correlation
values at discrete lags, which may be several dB lower than the prompt
correlator in a conventional receiver. This SNR loss may result in
undesired degradation in timing estimation. Furthermore, some issues
that have not been a problem to conventional GPS receivers arise
for a software receiver and one example is the sampling clock drift.
In this paper, we will introduce the concept of tracking GPS signal's
code phase and carrier frequency in the frequency domain. By ''frequency
domain'', we mean manipulating signal and correlation spectra to
achieve code and carrier alignment. After presenting a frequency
domain GPS receiver architecture, we will describe open loop and
closed-loop tracking mechanisms as well as computation-reduction
techniques for code and carrier estimation, which provide solutions
to the problems mentioned above. The proposed design has been successfully
tested with collected real GPS data and the processing results will
be presented to illustrate the concept and to substantiate the computational
algorithms.
Users
Please
log in to take part in the discussion (add own reviews or comments).