The progress in satellite navigation systems and other information and communication technologies enables the introduction of new services in the areas of road transport, merged under intelligent transport systems. Their primary purpose is to increase safety, efficiency and comfort, with positive effects for road users, transport infrastructure owners, and fleet managers.
The on-board unit in the vehicle, where individual applications are implemented, often combines the necessary navigation parameters from the GNSS receivers with motion sensors and a digital map to ensure a higher level of service quality and alleviate the problems that occur in certain difficult environments. In comparison with aviation, where the requirements for positioning quality are very high and the environment is relatively predictable, road environments constitute a set of varied and dynamic faulty impacts on positioning. This can lead to larger deviations in the navigation solution, which may not be a problem for some services, while others will exceed the expected limits.
In the first part of the dissertation, we focused on the types of road services with emphasis on critical services where the deviation of navigation parameters in the form of position, speed, direction of movement and exact time can lead to threats to safety, financial loss or regulatory violations. The overview includes advanced driver assistance services (ADAS), cooperative services (C-ITS), autonomous driving, electronic toll collection systems, insurance services, emergency service – eCall, digital tachographs and tracking of dangerous goods. We presented an expanded set of quality criteria for the navigation system and defined the basic criteria in the form of accuracy, availability and integrity, which are also the subject of deeper analyses. By comparing the available requirements of services from different types of traffic, we have found and demonstrated that currently there are no existing well-defined requirements in the road sector, which prevents the certification of the positioning module inside the vehicle. Therefore, we compared the two emerging standards in the field of GNSS-based positioning for road ITS, CEN EN 16803 and ETSI TS 103 246, where service architectures, metrics for performance assessment, and operational scenarios are defined. The CEN standard is additionally complemented with the results of the European COST Action SaPPART, where I actively participated in determining the service certification framework as well as the laboratory and field testing mechanisms, and in the classification of user location terminals for 4 years from this dissertation proposal.
The methodology of performing measurements within a Short-Term Scientific Mission (STSM) in September 2016 in the suburbs of Nantes, France, provides detailed aspects of the operational scenarios, measurement environments, the set of GNSS receivers and the reference positioning system. In accordance with EN 16803, the characteristics of four available environments, a city periphery, a motorway section, a city centre with medium-sized buildings and a rural area are presented. The installation of 3 groups with a total of 18 GNSS geodetic, car, and smartphone receivers with a dedicated Vehicle for Experimental Research on Trajectories (VERT) of the French Institute IFSTTAR are shown. After collecting all the measurements, we verified their validity and compliance with the existing data formats and showed common errors that may appear while performing similar campaigns using low-cost receivers.
The main part of the dissertation presents the results of an analysis of environmental impacts on the positioning integrity. Integrity, as a statistical measure of trust that can be placed in the correctness of the information supplied by the GNSS receiver, is addressed through the analyses of positioning accuracy changes due to the properties of satellite constellations, the signal propagation across the ionosphere and troposphere, the effects of the local environment on the obstructed visibility of satellites, reflections and attenuation of satellite signals, and the properties of receivers. We showed that the measurements position on the globe geometrically affects the expected accuracy through the DOP factor. From the 1.7 million samples captured, the statistical model of the correlation between the HDOP and the number of satellites was determined.
By using available daily broadcasted ephemeris from the web, we performed an evaluation of the number of not-received and received satellites. We proposed an innovative way of environment evaluation through the mathematical model, which results in the characteristic elevation angle, where the smoothed ratio between the numbers of not-received and received satellites reaches a certain percentage (5%, 10% and 15% respectively), or where the tangent to the ratio curve reaches the coefficient -1.
To determine the model of weights in the procedures for positioning and determining integrity, the C/N0 parameter was studied in detail, which in the output format NMEA of mass receivers provides the ratio between the power of the satellite carrier C and noise density N0 in 1 Hz bandwidth. We showed the distribution of the ratio in a clear-sky motorway and in obstructed urban environment for receivers with roof-placed antennas and smartphones behind the windshield, where urban influences cause an additional smaller peak of distribution about 20 dB lower than the main peakdue to the reception of attenuated reflected carriers. Due to the relation between the satellite elevation and its expected C/N0 ratio under clear-sky environment, a model of the template function was established, where, in the case of weights, the difference between the template and the measured value represents an additional weight factor in the exponential model. This model was verified with additional measurements on the roof of the Faculty of Electrical Engineering in duration of 24 hours, confirming the mathematical model of weights.
The contributions of individual receivers to accuracy were verified by the metric defined in the COST action SaPPART, based on the 50th, 75th, and 95th percentiles of the cumulative distribution function, and the numerical limits set from the technological and application perspectives. The initial hypothesis on visual gaps between geodetic, automotive and smartphone receivers on the CDF was proven incorrect during the measurement analysis. For example, the accuracy CDF of low-performance representatives of the automotive receivers group with roof antennas could not be distinguished from high-performance smartphones inside the vehicle. Thus, we also carried out the subgrouping of receivers according to their performances, which provided cleaner subgroup gaps.
This is followed by the classification of the receivers by horizontal, cross-track, and along-track accuracy, which demonstrates the high accuracy of geodetic receivers and low of smartphones, while the availability classification proves the opposite. Namely, smartphones try to provide a location with high availability at the cost of accuracy, while geodetic receivers provide high accuracy at the cost of availability. Automotive receivers thus represent a good balance between those two groups, as they ensure good accuracy and also a high degree of availability.
In the final content chapter, integrity monitoring is presented by weighting the measurements of the signals of individual satellites, where the significance of the weights in position estimation and protection level calculation is emphasized. Several methods for determining weights are considered: using satellites’ elevations, C/N0 signal ratios and combined methods, where an additional source of information, such as an electronic map or an open-sky detection camera is needed. For the sigma-ε model using the C/N0 ratios, a calibration procedure for the automotive receiver is presented across all four available environments. With calibrated weights, a comparison is made between the four different approaches of calculating the horizontal level of protection, which is evaluated in the closing part of the chapter with a practical test of the effectiveness of the weighted integrity model and representations in Stanford plots.