Precise point positioning (PPP) is a technique for processing Global Navi- gation Satellite Systems (GNSS) data, often using recursive estimation methods e.g. a Kalman Filter, that can achieve centimetric accuracies using a single receiver. PPP is now the dominant real-time application in o shore marine positioning industry. For high precision real-time applications it is necessary to use high rate orbit and clock corrections in addition to high rate observations. As Kalman filters require input of process and measurement noise statistics, not precisely known in practice, the filter is non-optimal. Geodetic quality control procedures as developed by Baarda in the 1960s are well established and their extension to GNSS is mature. This methodology, largely unchanged since the 1990s, is now being applied to processing techniques that estimate more parameters and utilise many more observations at higher rates. \Detection, Identification and Adaption" (DIA), developed from an optimal filter perspective and utilising Baarda's methodology, is a widely adopted GNSS quality control procedure. DIA utilises various test statistics, which require observation residuals and their variances. Correct derivation of the local test statistic requires residuals at a given epoch to be uncorrelated with those from previous epochs. It is shown that for a non-optimal filter the autocorrelations between observations at successive epochs are non-zero which has implications for proper application of DIA. Whilst less problematic for longer data sampling periods, high rate data using real-time PPP results in significant time correlations between residuals over short periods. It is possible to model time correlations in the residuals as an autoregressive process. Using the autoregressive parameters, the effect of time correlation in the residuals can be removed, creating so-called whitened residuals and their variances. Thus a whitened test statistic can be formed, that satisfies the preferred assumption of uncorrelated residuals over time. The effectiveness of this whitened test statistic and its impact on quality control is evaluated.

Real Time GNSS networks established across countries over the last fifteen years, provide centimetre level accuracy for a wide range of applications differing from precise ship docking to land surveying. Public sector organizations deploy GNSS Networks to support infrastructure projects. Private companies establish their own networks for commercial or private use. The number of users of these networks has steadily increased over the years, and a potential new market has been created. This thesis focuses on the exploitation of the advantages of using multiple users operating within a GNSS network as part of the system, as a virtual network of stations that can operate autonomously and combined with the reference station network. A new network-RTK methodology that encompasses the multiple users of network RTK services for instantaneous positioning, is presented. Users are equipped with a means of two-way communication that enables the data to be transmitted to a central processing facility on an epoch by epoch basis. A centralised approach reduces the need for complex algorithms at the user side. The methodology is tested through the use of data from the South California Integrated GNSS Network (SCIGN) for two different network formations with different numbers of users operating simultaneously. It is shown that regional estimation of relative ionospheric corrections from the proposed methodology has a positive effect on the overall ambiguity resolution success rates when compared to the use of a standard ionospheric model. It is demonstrated that the use of the multiple rovers operating in the network increase the ambiguity success rates for the ones outside the network area by almost 20%. A smaller improvement for rovers near the boundaries of the network is also achieved. Finally, new prospects for regional atmospheric modeling become available, since the algorithm estimates all the double difference ambiguities for the reference station-rover system.

Global Navigation Satellite Systems (GNSS) have become the positioning systems of choice for many applications. However, GNSS signals are susceptible to obstruction, interference and jamming. As a result, alternative technologies that can backup and operate independently of GNSS are required. One option is to use signals of opportunity, which are signals not designed or modi- ed speci cally for positioning. A key bene t is cost as there is no need to install and maintain additional transmission infrastructure. Furthermore, signals of opportunity have the extensive coverage and penetration necessary to reach di cult environments. The proposed positioning technique operates by comparing signals received at a user location with those received at a reference, or other user location. Signals are brought together and correlation-tested to obtain differential ranging measurements which are processed to yield the user's position. Hardware and signal processing solutions necessary to position with signals of opportunity are explored. The technology is validated on AM broadcast transmissions in the low frequency (LF) and medium frequency (MF) bands. Using these transmissions, results indicate that a positioning solution can be obtained in GNSS compromised environments, such as indoors and in urban canyons. In principle, the technique can be adapted to operate on different kinds of signals regardless of their modulation and content - so long as they can be separated from their counterparts. As a consequence of this, the concepts may be applied to other kinds of transmissions, with the potential of developing a system that achieves enhanced performance by exploiting heterogeneous transmissions from different parts of the spectrum.

Understanding travel behaviour and travel demand is of constant importance to transportation communities and agencies in every country. Nowadays, attempts have been made to automatically infer the modes of transport from positional data (such as GPS data) to significantly reduce the cost in time and budget of conventional travel diary surveys. Some limitations, however, exist in the literature, in aspects of data collection (spatio-temporal sample distribution, duration of study, granularity of data, device type), data pre-processing (managing GPS errors, choice of modes, trip information generalisation, data labelling strategy), the classification method used and the choice of variables used for classification, track segmentation methods used (clustering techniques), and using transport network datasets. Therefore, this research attempts to fully understand these aspects and their effect on the process of inference of mode of transport. Furthermore, this research aims to solve a classification problem of sparse GPS data into different transportation modes (car, walk, cycle, underground, train and bus). To address the data collection issues, we conduct studies that aim to identify a representative sample distribution, study duration, and data collection rate that best suits the purpose of this study. As for the data pre-processing issues, we standardise guidelines for managing GPS errors and the required level of detail of the collected trip information. We also develop an online WebGIS-based travel diary that allows users to view, edit, and validate their track information to assure obtaining high quality information. After addressing the validation issues, we develop an inference framework to detect the mode of transport from the collected data. We first study the variables that could contribute positively to this classification, and statistically quantify their discriminatory power using ANOVA analysis. We then introduce a novel approach to carry out this inference using a framework based on Support Vector Machines (SVMs) classification. The classification process is followed by a segmentation phase that identifies stops, change points and indoor activity in GPS tracks using an innovative trajectory clustering technique developed for this purpose. The final phase of the framework develops a network matching technique that verifies the classification and segmentation results by testing their obedience to rules and restrictions of different transport networks. The framework is tested using coarse-grained GPS data, which has been avoided in previous studies, achieving almost 90% accuracy with a Kappa statistic reflecting almost perfect agreement.

The advent of global navigation satellite systems, and especially GPS, signified a new era in navigation. Although GPS tends to be a panacea in terms of positioning and tracking, there are still cases whereby severe problems can render navigation virtually impossible. Difficult GNSS environments such as forests, heavy canopy covered areas, or urban canyons can have a negative impact on the propagation of satellite signal and introduce large errors in positioning due to signal attenuation. Several techniques have been developed to deal with this issue, e.g. integrated GNSS/INS navigation configurations. This thesis focuses on the application of terrain aiding and its impact on the availability, accuracy, and reliability of GNSS positioning. Furthermore, it examines the impact of using GNSS receivers of different grade and different chipset sensitivity in such operations. Two main experiments were carried out; the first was conducted in the mountains of Transylvania, in Romania. This project was sponsored by the European GNSS supervisory authority, and involved the tracking of animals (bears in this case). The second experiment was held in Greenwich Park, London, using an accurate terrain model, kindly provided by the Ordnance Survey of Great Britain. In both cases, new algorithms have been developed to combine satellite tracking with local terrain models to improve positioning performance. The results indicate that in the case of standard sensitivity receivers the use of an accurate digital terrain model can improve positioning availability by as much as 74%, and the accuracy of normal four-satellite positioning by a factor of up to five. Furthermore, it was found that, for both standard and high sensitivity receivers, the external reliability of position fixes can be improved by an average of 40%, and up to 90% in some cases.

Among the various factors limiting accurate positioning with a Global Navigation Satellite System (GNSS) is the inherent code error level on a code observation, cycle slip occurrence on a phase observation, inadequate accuracy in the broadcast ionospheric model for single-frequency receivers; and the occurrence of observation gaps, which are short duration satellite outages (temporal loss of an observed satellite). The existing Cycle Slip Detection and Correction (CSDC) techniques are usually multi-satellite based; quite computationally intensive; and are often marred by the inherent code errors from the included code observations. Also, existing code-carrier smoothing techniques employed to mitigate code errors are limited by cycle slip occurrences on phase observations. In this research, algorithms are proposed in order to facilitate simple, efficient and real-time cycle slip detection, determination and correction, on a standalone single- or dual-frequency receiver; to enable cycle-slip-resilient code errors mitigation; and to improve the broadcast ionospheric model for single-frequency receivers. The proposed single-satellite and phase-only-derived CSDC algorithms are based on adaptive time differencing of short time series phase observables. To further provide robustness to the impact of an observation gap occurrence for an observed satellite, post-gap ionospheric delay is predicted assuming a linearly varying ionospheric delay over a short interval, which consequently enables the dual-frequency post-gap cycle slip determination and code error mitigation. The proposed CSDC algorithms showed good performance, with or without simulated cycle slips on actual data obtained with static and kinematic GNSS receivers. Over different simulated cycle slip conditions, a minimum of 97.3% correct detection and 79.8% correctly fixed cycle slips were achieved with single-frequency data; while a minimum of 99.9% correct detection and 95.1% correctly fixed cycle slips were achieved with dual-frequency data. The point positioning results obtained with the proposed methods that integrates the new code error mitigation and cycle slip detection and correction algorithms, showed significant improvement over the conventional code-carrier smoothing technique (i.e. a standalone Hatch filter, without inclusion of any cycle slip fixing method). Under different simulated cycle slip scenarios, the new methods achieved 25-42% single-frequency positioning accuracy improvement over the standalone Hatch filter, and achieved 18-55% dual-frequency positioning accuracy improvement over the standalone Hatch filter.

In order to obtain high quality positions from navigation satellites, range errors have to be identified and either modelled or estimated. This thesis focuses on satellite clock errors, which are needed to be known because satellite clocks are not perfectly synchronised with navigation system time. A new approach, invented at UCL, for the simultaneous estimation, in a single epoch, of all satellite clock offsets within a Global Navigation Satellite System (GNSS) from range data collected at a large number of globally distributed ground stations is presented. The method was originally tested using only data from a limited number of GPS satellites and ground stations. In this work a total of 50 globally distributed stations and the whole GPS constellation are used in order to investigate more fully the capabilities of the method, in terms of both accuracy and reliability. A number of different estimation models have been tested. These include those with different weighting schemes, those with and without tropospheric bias parameters and those that include assumptions regarding prior knowledge of satellite orbits. In all cases conclusions have been drawn based on formal error propagation theory. Accuracy has been assessed largely through the sizes of the predicted satellite clock standard deviations and, in case of simultaneously estimating satellite positions, their error ellipsoids. Both internal and external reliability have been assessed as there are important contributions to integrity, something that is essential for many practical applications. It has been found that the accuracy and reliability of satellite clock offsets are functions of the number of known ground station clocks and distance from them, quality of orbits and quality of range measurement. Also the introduction of tropospheric zenith delay parameters into the model reduced both accuracy and reliability by amounts depending on satellite elevation angles.

In an operational sense, satellite clock time offset prediction (SCTOP) is a fundamental requirement in global navigation satellite systems (GNSS) tech- nology. SCTOP uncertainty is a significant component of the uncertainty budget of the basic GNSS pseudorange measurements used in standard (i.e not high-precision), single-receiver applications. In real-time, this prediction uncertainty contributes directly to GNSS-based positioning, navigation and timing (PNT) uncertainty. In short, GNSS performance in intrinsically linked to satellite clock predictability. Now, satellite clock predictability is affected by two factors: (i) the clock itself (i.e. the oscillator, the frequency standard etc.) and (ii) the prediction algorithm. This research focuses on aspects of the latter. Using satellite clock data—spanning across several years, corresponding to multiple systems (GPS and GLONASS) and derived from real measurements— this thesis first presents the results of a detailed study into the characteristics of GNSS satellite clocks. This leads onto the development of strategies for modelling and estimating the time-offset of those clocks from system time better, with the final aim of predicting those offsets better. The satellite clock prediction scheme of the International GNSS Service (IGS) is analysed, and the results of this prediction scheme are used to evaluate the performance of new methods developed herein. The research presented in this thesis makes a contribution to knowledge in each of the areas of characterisation, modelling and prediction of GNSS satellite clocks. Regarding characterisation of GNSS satellite clocks, the space-borne clocks of GPS and GLONASS are studied. In terms of frequency stability—and thus predictability—it is generally the case that the GPS clocks out-perform GLONASS clocks at prediction lengths ranging from several minutes up to one day ahead. There are three features in the GPS clocks—linear frequency drift, periodic signals and and complex underlying noise processes—that are not observable in the GLONASS clocks. The standard clock model does not capture these features. This study shows that better prediction accuracy can be obtained by an extension to the standard clock model. The results of the characterisation and modelling study are combined in a Kalman filter framework, set up to output satellite clock predictions at a range of prediction intervals. In this part of the study, only GPS satellite clocks are considered. In most, but not all cases, the developed prediction method out- performs the IGS prediction scheme, by between 10% to 30%. The magnitude of the improvement is mainly dependent upon clock type.

Precise Point Positioning (PPP) provides GNSS navigation using a stand-alone receiver with no base station. As a technique PPP suffers from long convergence times and quality degradation during periods of poor satellite visibility or geometry. Many applications require reliable realtime centimetre level positioning with worldwide coverage, and a short initialisation time. To achieve these goals, this thesis considers the use of GLONASS in conjunction with GPS in kinematic PPP. This increases the number of satellites visible to the receiver, improving the geometry of the visible satellite constellation. To assess the impact of using GLONASS with PPP, it was necessary to build a real time mode PPP program. pppncl was constructed using a combination of Fortran and Python to be capable of processing GNSS observations with precise satellite ephemeris data in the standardised RINEX and SP3 formats respectively. pppncl was validated in GPS mode using both static sites and kinematic datasets. In GPS only mode, one sigma accuracy of 6.4mm and 13mm in the horizontal and vertical respectively for 24h static positioning was seen. Kinematic horizontal and vertical accuracies of 21mm and 33mm were demonstrated. pppncl was extended to assess the impact of using GLONASS observations in addition to GPS in static and kinematic PPP. Using ESA and Veripos Apex G2 satellite orbit and clock products, the average time until 10cm 1D static accuracy was achieved, over a range of globally distributed sites, was seen to reduce by up to 47%. Kinematic positioning was tested for different modes of transport using real world datasets. GPS/GLONAS SPPP reduced the convergence time to decimetre accuracy by up to a factor of three. Positioning was seen to be more robust in comparison to GPS only PPP, primarily due to cycle slips not being present on both satellite systems on the occasions when they occurred, and the reduced impact of undetected outliers.

The International Maritime Organisation (IMO) established a set of positioning requirements for future Global Navigation Satellite System (GNSS) constellations in IMO resolution A.915. It is important to be able to determine if these requirements can be met, and what shore infrastructure would be required. This thesis describes the collection of data in a marine environment and the analysis of these data with regards to the requirements. The data collection exercise was held at the beginning of May 2008 and saw THV Alert navigate into Harwich Harbour whilst Global Positioning System (GPS) observation data were recorded from onboard the vessel and from shore-based reference stations. Additional data were obtained from nearby Ordnance Survey reference stations, and two total stations were used to track the vessel’s passage to provide a truth model. Several modernised GPS satellites were tracked. The data were processed under different scenarios, using software developed at UCL, and the positioning performance was analysed in the context of the IMO requirements. Potential performance improvements from modernised GPS and Galileo were then discussed. Providing integrity through single-epoch real-time kinematic positioning, required to meet the strictest IMO requirements, is particularly difficult. The identification of phase observation outliers is not possible before the integer ambiguities are resolved, but an undetected outlier could prevent successful ambiguity resolution. It will not always be necessary to fix all the ambiguities to achieve the required positioning precision, particularly with a multi-GNSS constellation. This thesis introduces a new algorithm for partial ambiguity resolution in the presence of measurement bias. Although computationally intensive, this algorithm significantly improves the ambiguity resolution success rate, increasing the maximum baseline length over which the highest requirements are met with dual-frequency GPS from 1 km to 66 km.