Harmonization of NPRS Observations for a Seamless RTK Positioning Service in Automated Driving Applications

Abstract

The application of GNSS for positioning in automated driving applications usually requires utilizing both code and carrier phase measurements, together with augmentation techniques such as Real Time Kinematic (RTK), Precise Point Positioning (PPP) or combinations thereof. The corrections can be distributed in different ways, via satellite communication channels or terrestrial communication channels. The reference data is generated based on measurements by networks of reference stations, typically in a local reference frame. Public or private entities offer services on top of these networks, often covering large areas. The Swedish SWEPOS network for example offers Network RTK (OSR) nationwide and is commonly used as a tool for land surveying. The network corrections are commonly encoded using standards published by the Radio Technical Commission for Maritime Services (RTCM) and distributed as reference data using Networked Transport of RTCM via Internet Protocol (NTRIP). Recently, the 3rd Generation Partnership Project (3GPP), responsible for mobile communication specifications, has specified encoding of NRTK data based on RTCM as part of the LTE Positioning Protocol (LPP). Such data can either be provided via unicast or via cellular broadcast by a mobile operator to allow scalability to support mass-market deployment. With a typical 35 km spacing of physical reference stations (PRS), an NRTK device can achieve centimeter level positioning accuracies using integer ambiguity resolution, thus well meeting the requirements of automated and autonomous driving. However, to provide augmentation to mass-market applications a typical NRTK service faces several problems, such as the scaling of communication and computation. Also, the kinematic user needs to organize the handling of reference station changes along the path she or he is traveling. The required integer ambiguity resolution can be a time-consuming process, especially in complicated environments. This can affect the availability of high accuracy position estimates, which in turn can be problematic for highly dynamic, safety critical scenarios. Ambiguity differences between reference stations can be handled in different ways on the user side but may also be supported by the network. This paper suggests methods for harmonizing the integer phase ambiguity for an entire network such that users may rely on coherent augmenting observations during the changes of reference stations over large areas. Data streams from different PRS exhibit differences in the ambiguities of the carrier phase observations, which is dependent on the receiver types and methods of alignment to code observations. Modern installations are usually erroneous by a few cycles but may deviate considerably. Non-physical reference stations (NPRS or virtual reference stations VRS), which are calculated by using several PRS, do propagate these cycle errors and may further introduce different ambiguities. Ideally, the NRTK software of the service provider could handle coherent carrier phase ambiguities, but the current demands on those systems do not require this functionality and is thus simply not present. An extra computational layer between the service and the user can be used to provide harmonized observations. Considering a grid of reference stations transmitting to a location server for distribution, the commonly used RTCM transport is intercepted and monitored. An ambiguity resolution of the grid results into a set of integers, which are used to align the phase observations. As an additional feature, this processing can introduce another layer of abstraction by referencing the positions of several reference stations in a wide area to a single joint location that makes changes of references transparent to the user, a concept named virtual virtual reference station (VVRS). The intermitted processing may also be used to introduce a traceable common network clock into the observations, which can be explored by applications requiring precision timing. The aligned reference data is repacked into RTCM streams and then distributed to the location server. To correctly solve ambiguities and clocks of a large network can be non-trivial and needs strategies for the propagation of fixed ambiguity relationships through the network, managing the introduction and removal of satellites as they rise and fall, separation detection and holes. It further requires means of communicating the ambiguity state of the distributed reference data. One possibility is to use the integer ambiguity level indications that are introduced in 3GPP LPP, which is open and fully interoperable. This paper exemplifies a network-based solution of carrier phase integer ambiguity harmonization using the SWEPOS network using a limited number of reference stations. We also show the improved RTK performance of harmonized network observations on a stretch of highway between the Swedish cities of Gothenburg and Borås. Work present in this paper is supported by the NPAD project, Network-RTK Positioning for Automated Driving. NPAD is a Swedish research project sponsored by Vinnova - FFI. It targets research for connected vehicles with the aim to provide absolute positioning using the national reference network SWEPOS and NRTK distributed via 3GPP LPP Rel 15 and 16. (https://www.diva-portal.org/smash/get/diva2:1530169/FULLTEXT01.pdf )