Resource-efficient parallel acquisition architectures for modernized GNSS signals

Abstract

Abstract The acquisition of global navigation satellite system (GNSS) signals is an extremely computa-tionally intensive task. This explains that the first GPS receivers needed a very long time toobtain a position. Thanks to technological advances, it is now possible to use highly parallelimplementations and efficient algorithms based on the fast Fourier transform (FFT), and thusreduce significantly the processing time. Indeed, today, it takes less than a second to detect aGPS L1 C/A signal with a clear view of the sky. However, this case does not correspond to allthe situations, and does not mean that the research on this topic is completed. For example, itis always necessary to reduce the power consumption of GNSS devices embedded in portableelectronics equipment to further improve their battery autonomy. Sometimes, it is necessaryto detect very weak signals, such as in space applications because of the long distances andthe bad geometry, or with receivers embedded in smartphones where the antenna must meetaesthetic criteria, which leads to poor performance in terms of gain. And finally, the recentintroduction of new GNSS signals will lead to better performance, but at the same time willrequire a much more complex signal processing. Therefore, it is still necessary to find newalgorithms in order to meet the current needs of the society and scientists.The goal of this Ph.D. thesis has been to search algorithms to reduce the complexity of theacquisition, in order to reduce the processing time or the resources used, depending onthe context. The research has focused on the computation of the correlation using FFTs,and on reducing its complexity by exploiting the characteristics of the GNSS signals. Thisresearch has been performed with a hardware implementation in mind rather than a softwareimplementation, because this Ph.D. thesis started with a project where the goal was to developa GPS receiver on a programmable circuit (more specifically on a field programmable gatearray, or FPGA).In this thesis, first we show simple methods to reduce the complexity of the FFT or of acorrelation computed by FFT (or of a convolution) on Altera FPGAs. In particular the methodswe propose allow a significant reduction of the memory resources. Moreover, the applicationof these methods is not restricted to GNSS signals, but in fact apply to any other systemscomputing FFTs, convolutions, or correlations, since no assumption is made on the signals.Afterwards, we focus on the acquisition of modern signals having a secondary code using theparallel code search, and especially on the acquisition of the GPS L5 signal. Two cases areconsidered :vii