Book: Navigation Signal Processing for GNSS Software Receivers (Gnss Technology and Applications)
Publisher: Artech House Publishers
The continuous developments of software-defined radio technology resulted in the appearance of the first real-time GPS software radios at the beginning of this century. For the first time, it was possible to realize a complete GNSS receiver without going into the depths of cumbersome hardware development that requires development or programming of low-level digital circuitry. The hardware development efforts were indeed so high that only a very limited number of companies or research institutes could afford them. Furthermore, the implementation constraints were so severe, especially for the first generation of GPS receivers, that often crude signalprocessing approximations were necessary to allow a real implementation. Currently, software-defined radio technology not only allows receiver implementations by a larger research community, but also drastically increases the signal-processing capabilities. It also has the potential to become, in certain navigation areas, a commercial success.
Software radio technology provides an opportunity to design a new class of GNSS receivers, being more flexible and easier to develop than their FPGA- or ASIC-based counterparts. Therefore, this text reviews navigation signal detection and estimation algorithms and their implementation in a software radio. A focus is put on high-precision applications for GNSS signals and an innovative RTK receiver concept based on difference correlators is proposed.
This text makes extensive use of the least-squares principle. The least-squares principle is the typical basis for the calculation of a navigation solution. An adjustment or a Kalman filter calculates positions from pseudorange observations in virtually any GNSS receiver. Within this text, the least-squares principle is consistently extended to also allow signal samples as observations. In contrast to the pseudorangeobservation equation, the signal sample model is highly non-linear, causing a number of difficulties that are discussed. Furthermore, signal sample observations can be complex-valued.