The objective of this activity was to provide a study on algorithms for achieving real-time, high-precision positioning with mass-market GNSS receivers for automotive applications in urban environments and highways. The main output of the work was required to be a prototype demonstrator, consisting of a software implementation of a selected set of techniques/architectures that could be implemented in real mass-market GNSS receivers for positioning performance improvement in the near future.
In order to achieve these objectives, NSL has enhanced their existing MSP3 Precise Point Positioning (PPP) software. Rather than taking the classical PPP approach of forming ionospherically-free observables (linear combination), MSP3 instead takes an ‘undifferenced’ approach, keeping measurements on different frequencies separate. As a baseline it involves the optional use of ionospheric pseudo-observations which provide absolute constraints on the slant ionospheric delay states to be estimated.
Enhancements made to MSP3 have included a) taking into account the characteristics of mass market receivers including measurement noise, clock behaviour etc; b) adding GLONASS and Beidou functionality to GPS and Galileo, as already implemented; c) enhancing pre-processing algorithms e.g. cycle slip detection; d) improving the PPP Kalman filter including dynamic model, measurement weighting method and filter RAIM; e) adding the use of complementary sensors (gyroscope and odometer); f) adding Integer Ambiguity Resolution (IAR) capability; and g) introducing algorithms to handle low numbers of satellites and multipath effects in extreme urban environments. In addition, a version of the software has been generated that can process data in real-time, including obtaining RTCM messages over NTRIP.
In the first phase of the project, points a) to e) were implemented. Real GNSS and sensor data was collected in the UK and China (in order to provide Beidou data) and truth trajectories generated. The software was then subjected to a thorough test campaign, analysing the sensitivity of performance to parameters including receiver type, numbers of constellations, orbit and clock data, use of complementary sensors, ionosphere data and bias data. The enhanced MSP3 PPP engine has been found to be able to achieve sub-metre (several tens of centimetres) horizontal positioning accuracy at the 50th and 68th percentile levels in both rural and urban areas. Fully continuous positioning (100% availability) is possible in urban areas when complementary sensors are used. The use of multiple constellations was found to provide a significant performance benefit in urban environments.
The second phase of the project has involved implementing points f) and g), as well as the real-time software. The impact on performance of these features is currently being evaluated. The added value of increased Galileo satellite availability will also be assessed by using new data, collected by ESA.

The next generation of low-cost, multi-frequency, multi-constellation GNSS receivers, boards, chips and antennas are now quickly entering the market. These sensors are offering to disrupt portions of the precise GNSS positioning industry with much lower cost hardware and promising to provide precise positioning to a wide range of consumers. The presented work provides a timely, thorough investigation into the potential disruption and promise. A systematic and rigorous set of experiments have been carried-out, collecting measurements from a wide array of low-cost, dual-frequency, multi-constellation GNSS boards, chips and antennas introduced in late 2018 and early 2019. These sensors range from dual-frequency, multi-constellation chips in smartphones to stand-alone chips and boards. In order to be comprehensive and realistic, these static and kinematic experiments were conducted in benign, typical, suburban and urban environments.

In terms of processing the raw measurements from these sensors, the Precise Point Positioning (PPP) GNSS measurement processing mode has been used. PPP has become the defacto GNSS positioning and navigation processing technique for scientific and engineering applications that require dm- to cm-level positioning in remote areas with few obstructions and provides for very efficient worldwide, wide-array augmentation corrections. While real-time kinematic (RTK) and network RTK dominate urban and suburban markets, it was deemed of great scientific interest to assess the PPP performance with these hardware options with and without additional local augmentation corrections.

Analysis of the raw measurements illustrates a) the lower signal availability and b) the weaker signal strength from the chip/antenna combinations as compared to geodetic quality instrumentation, c) high pseudorange multipath and noise, and d) some interesting measurement curiosities. While there are significant measurement gaps in some datasets, most carrier-phase measurement streams are of low noise with limited cycle slips. Significant customization to the York-PPP processing engine was necessary, especially in the quality control and residual analysis functions, in order to successfully process these datasets. Results for new smartphone sensors show positioning performance is typically at the few dm-level with limited convergence period – 1 to 2 orders of magnitude better than standard point positioning. The GNSS chips and boards combined with higher-quality antennas produce positioning performance approaching geodetic quality. And under ideal conditions, phase ambiguities are resolvable. However, there are a number of caveats to these performance assessments, as consistence of results is lower for all of these new sensors as compared to geodetic hardware.

These results are very promising for the use of PPP and RTK in next-generation GNSS sensors for smartphone, vehicle, Internet of things (IoT), etc. applications. Future work includes further tuning of the measurement processing, ambiguity resolution, and use of AI techniques for measurement filtering.

The current constellation of Galileo satellites consists of 26 space vehicles, among which are 2 unavailable satellites and 2 satellites in elliptical orbits. For the remaining 22 healthy satellites there are not only broadcast ephemeris available, but also precise products are provided by the International GNSS Service (IGS) under the Multi-GNSS Experiment (MGEX) and real-time products are transmitted by the Centre National d'Études Spatiales (CNES). Therefore, multi-GNSS solutions can be computed, but also Galileo-only positioning is feasible, which even includes absolute positioning in real-time and post-processing mode.
In this study, the global availability of Galileo satellites is analysed by means of the number of visible satellites and corresponding PDOP for a variety of elevation angles. Moreover, a comprehensive evaluation of dual-frequency absolute positioning using Galileo is presented and the performance is compared to the one of GPS positioning. We evaluate the performance of daily static and kinematic positioning using pseudoranges only (SPP) as well as pseudoranges with carrier phase data (PPP), based on broadcast ephemeris, real-time CNES products and final MGEX products.
We found in more than 99.9% of the time at least 5 Galileo satellites are in view above 10 degree elevation. The performance of Galileo PPP is still not as good as the one of GPS, due to the lower accuracy of final and real-time orbit and clock products. However, an outstanding accuracy of Galileo-only positioning based on broadcast ephemeris is confirmed. The RMSE of static Galileo SPP is 0.14 and 0.43 m for the horizontal and vertical components, respectively, which is almost 2 times better than GPS. Moreover, sub-decimetre accuracy of daily static Galileo PPP based on broadcast ephemeris is achieved. The performance of Galileo-only and GPS-only kinematic positioning is similar, except for PPP analysis using real-time products, which is more accurate for GPS. Finally, we notice that SPP solutions, for both GPS and Galileo, benefit from real-time products, leading to an increase of the accuracy by up to 59%.

Several experiments using GNSS real time technology on board of a survey vessel have been conducted in the last three years by the Federal Agency for Cartography and Geodesy (BKG) during the EU co-financed project FAMOS odin. The focus of the investigations was on testing of precise point positioning (PPP) technique together with state space representation (SSR) corrections under kinematic conditions. During three 10-days field campaigns different SSR correction data streams were used and positioning solutions are compared to “classical” real time kinematic (RTK) techniques as well as post-processing solutions.

Currently, the German surveying vessels use positioning services, which are based on the RTK approach. Here, corrections are computed individually for each user, which requires a bidirectional communication link between the user and the positioning service. However, due to this and the geometry of the reference station network, this service is not or not fully available far off the coast. The alternative PPP technology is based on the SSR approach. For this approach, a unidirectional communication link is sufficient and allows accurate positioning in remote regions resp. the open seas.

The SSR correction data are calculated and provided in open data formats via internet to the user. The latency of such products is one key parameter, which may limit the usability of those products significantly. This is especially true for applications with a high requirement on the time line, such as kinematic positioning. The latency optimization as well as the identification of single points of failure of the entire production chain require a detailed analysis of every single step. Depending on whether a combination of the individual SSR contributions is included or not, the total delay may sum up to 10 to 25 seconds.

A further critical element to get a robust production chain is the redundancy of data transportation from the single station of a GNSS reference network to the analysis centers and finally to the user. This is especially critical if the user is outside a well-covered communication network, for example on a vessel far away from the coast. The quality of the service has to be a well-balanced combination of system robustness and the timeline of the complete process chain. In order to analyse all these different aspects, BKG can act as service provider for SSR correction as well as a service user using the BKG Ntrip Client, which allows PPP based on own or third-party SSR corrections.

After the successful tests mentioned above, BKG is going to implement a permanent installation of the PPP technique on one survey vessel in autumn 2019 in order to set up all different elements of the real-time GNSS processing chain in an operational level. The paper presents the various levels of criticality concerning service robustness and related main aspects of the implementation. First experiences of the vessel’s crew with the operational chain and first results of the achieved accuracy in positioning will be also presented.