High-Precision GNSS Kinematic Relative Positioning: Methods and Assessment

Abstract

In today's dynamic positioning applications, achieving sub-meter or centimeter-level precision has become increasingly crucial in various kinematic positioning applications. This holds particularly true for navigating rovers within challenging environments where GNSS measurements' quality limitations hinder accurate positioning. To enhance positioning accuracy, GNSS carrier phase measurements have gained significant importance. These measurements provide tracking precision approximately a hundred times finer than pseudorange measurements. However, these measurements present a challenge known as 'integer ambiguity,' which prevents their direct application in positioning. While a portion of the carrier phase measurement can be accurately measured, the remaining portion, representing a full cycle, remains unknown. Resolving this unknown cycle number is known as the 'ambiguity estimation problem' within the context of GNSS positioning. Achieving centimeter-level positioning accuracy is only possible if we successfully resolve this integer cycle number.

This dissertation focuses on applying carrier phase measurements in demanding scenarios, particularly in vehicle positioning. Our primary goal is to develop an efficient algorithm and software for achieving centimeter-level accuracy in kinematic positioning using carrier phase measurements in these challenging contexts. We realize this goal through GNSS differential techniques that mitigate errors in GNSS-derived positions by leveraging correlated errors in GNSS measurements, augmented with data from a reference GNSS receiver.

The dissertation explores various ambiguity resolution methods, including Full Ambiguity Resolution (FAR), Partial Ambiguity Resolution (PAR), and Ionosphere-free (Iono-free), aimed at improving position determination accuracy and enabling the reliable use of carrier phase measurements. We implement MATLAB software to enhance GNSS positioning accuracy, utilizing an extended Kalman filtering technique to better model the kinematics of a moving receiver.

Our assessment is based on real-world data collected from Ohio State University's west campus area and CORS stations, spanning diverse baseline lengths and receiver types, from geodetic to low-cost receivers. Our software extends to multi-baseline scenarios, surpassing single baselines.

Notably, the PAR method significantly enhances solution fix rates compared to the FAR and Iono-free methods, especially in short baseline scenarios, resulting in improved positioning accuracy. The comprehensive testing and validation conducted underscore the substantial contribution of our software to achieving centimeter-level kinematic positioning accuracy.

Moreover, the analysis demonstrates that with the increase in baseline lengths, FAR and PAR methods exhibit a decline in fix rates and an increase in standard deviation values. This highlights the challenge of maintaining precise and fixed solutions over longer baselines. The PAR method outperformed the FAR method, especially in terms of fix rates, particularly for longer baselines. On the other hand, the Iono-free method faced significant challenges in high-dynamic scenarios with longer baselines, resulting in decreased precision and reliability.

For CORS station data only in the case of medium to long baseline lengths, the Iono-free method exhibited superior performance compared to the FAR and PAR methods. However, the PAR method delivered results on par with those obtained from the Iono-free method. In contrast, the FAR method faced challenges in maintaining the same level of performance as the PAR and Iono-free methods for these longer baseline scenarios.