On dynamic array processing for GNSS software receivers

This thesis presents contributions in the field of satellite navigation with a focus on array processing and related implementation issues. For readers not familiar with GNSS, it also includes a brief overview of satellite navigation.Compared to the state of the art only ten years ago, modern GNSS receivers are very capable. One reason for this improvement is advances in the semiconductor industry that have increased both the available processing power and the energy efficiency. An active research community have also made important contributions resulting in more sophisticated algorithms. To improve receiver performance even further, auxiliary sensors such as gyros and accelerometers are becoming increasingly common. A related option involves using an antenna with several physical elements. This is known as an antenna array and is often used for radar, sonar and telecommunication applications. Array processing can also be used for GNSS and as such it is the primary focus of this thesis. An array allows for exploration of the spatial domain, in other words a receiver that may differentiate between signals depending on the direction of arrival. For GNSS, where interference and multipath (signal reflection off, for example, buildings or the ground) may be significant sources of error, this is an attractive solution. Although array processing have been the subject of extensive research efforts within other fields, there are several remaining issues with regards to how these techniques can be implemented in a GNSS receiver.With regards to array processing there are also properties unique to GNSS, such as multiple signal sources at known positions, that have not been explored sufficiently in previous efforts. In this thesis we show how these properties can be exploited to improve receiver performance in dynamic scenarios. In short, the orientation of the antenna platform is estimated accurately (typical variance around 1°) using beamforming techniques. This information is then used to achieve a better estimate of the radio environment by allowing for longer integration periods when estimating the covariance matrices. A better estimate of the covariance matrices directly translates into improved receiver performance, especially so in areas of moderate levels of multipath/interference.Further, a method to calibrate GNSS array antennas using real signals is investigated in detail. Instead of resorting to electromagnetic simulations that requires precise knowledge about the antenna and installation factors, or RF chamber measurement that is expensive, it is shown how the array antenna can be calibrated using live signals. The accuracy of the resulting model is verified using real data.Also, the first implementation of an RF record and replay system is presented. With such a system data can be recorded in a specific environment, generally a time consuming task, and later played back into the antenna input of any GNSS receiver. Such systems are nowadays commercially available and have proven very useful for testing and validation of GNSS receivers. Throughout the thesis, the required receiver architecture and practical viability of the proposed algorithms are considered