| Summary: | In the past, in order for GPS (Global Positioning System) to work accurately, the presence of an unobstructed LOS (Line-Of- ight) signal was necessary. Weak signals were not suitable for use because they may have large associated noise and other errors. The expansion of GPS to LBS (Location- ased Services) and other navigation applications all over the world, such as the E-911 and the E-112 mandates in the United States and Europe respectively, changed the paradigm. Consequently a dramatic increase in the need for more and more performant positioning techniques is expected, especially in urban and indoor environments. These rising localization requirements pose a particularly difficult challenge for GPS receivers design. The thesis objective is to evaluate and enhance existing GPS signal acquisition techniques for positioning goals in harsh environments, in the context of AGPS (Assisted GPS). The AGPS system assumes that the GPS receiver is connected to or introduced in a mobile phone. This allows for the transfer of AD (Assistance Data) to the GPS receiver via the GSM (Global System for Mobile communications) cellular network. Amongst others, the AD provides the GPS receiver with the list of visible satellites and estimates of their Dopplers and code delays, thus reducing the search window of these parameters. This work consists in exploring different GPS signal acquisition to reduce the acquisition time or TTFF (Time To First Fix), without affecting the receiver sensitivity. This is done after a prior study of the GPS radio channel. The study starts out with a revue of the GPS system and the GPS transmitted and received signal structure. The acquisition process is then described in details: the classical acquisition is first described in order to proceed afterwards with the impact of the propagation environment on this stage of the signal processing. For this purpose, harsh environments (urban and indoor) are modelled and analysed. This analysis enables to study the problems which encounter the radio frequency signal propagation through such environments. Note that the urban channel is studied using an existing statistical model developed by Alexander Steingass and Andreas Lehner at the DLR (German Aerospace Center) [Steingass et al., 2005]. On the other hand, an indoor channel model was developed by the ESA (European Space Agency) in the frame of a project entitled “Navigation signal measurement campaign for critical environments” and presented in [Pérez-Fontán et al, 2004]. But this model considers a time invariant statistical channel. Consequently, we developed an Indoor model
which rather considers a time variant channel, by taking into account temporal variations of
some channel parameters, like the transfer function delay and phase. The initial values are
however based on the statistical distributions provided by the ESA model. The channels are analysed is terms of multipaths, cross-correlations and signal masking. The multipaths replicas are particularly disturbing in urban environments while the cross-correlations and masking effects are more disturbing in indoor environments. These phenomena may induce errors in the final solution calculated by the receiver. In order to avoid this error, one solution consists in increasing the signal observation duration in order to enhance the signal to noise ratio. But this generally implies longer acquisition time, thus affecting the receiver iv performance, commercially speaking. Indeed, the time requirements are as important as sensitivity requirements for GPS users. However, these two requirements are not generally compatible with each other. Consequently, an ideal solution consists in reducing the acquisition time without greatly affecting the receiver sensitivity. Accordingly, such advanced methods for acquisition signal processing are described next. Most of these methods aim at reducing the total acquisition time, rather than enhancing the receiver sensitivity. This means however that longer signal blocks can be processed (thus enhancing sensitivity) without affecting the global processing duration. At first, each of these methods is evaluated through the description of its advantages and drawbacks. A performance evaluation of these algorithms, using signals generated with a Spirent STR4500, ensues as a final step of this study
|
|---|
