July 23, 2020 – By Matteo Luccio
Thirty years ago, more than a decade before most people had even heard of GPS, receiver manufacturers and surveyors were already improving by providing and using correction services to compensate for errors in the system—including clock drift, orbit errors, signal errors, atmospheric errors and multipath.
Today, dozens of public and commercial correction services enable users to achieve accuracies of decimeters, centimeters or even millimeters. Also, many GNSS processing services correct measurements taken in the field using data from reference points. Increasing positioning accuracy has become the cornerstone of modern GNSS practice.
The current boom for correction services is driven mostly by the demand for high accuracy from the automotive industry (including the development of self-driving cars), as well as smart consumer devices and various forms of automation. Automotive companies and telecoms are deploying infrastructure around the globe to provide centimeter-level positioning. GNSS satellites also can transmit corrections directly, as the Japanese CLAS service from the QZSS constellation does, and Galileo’s High-Accuracy Service (HAS) soon will. To compensate for receiver-side issues — multipath, jamming and spoofing — some GNSS receivers also incorporate advanced positioning algorithms.
Clock and orbit errors are specific to each satellite; they do not depend on the position of the receiver. But atmospheric errors are introduced when the signal travels from the satellites to the user. Reference stations (base stations) of GNSS receivers installed at fixed and precisely surveyed positions provide corrections that compensate for both sets of errors to the rovers carried by field crews. When connected, reference receivers spread over a geographic area form reference networks, such as that of continuously operating reference stations (CORS). Achieving maximum accuracy requires initializing the receiver, which can take a few seconds to several minutes, depending on the type of corrections.
Established and new methods
Two established methods have been used for decades.
Real-time kinematic (RTK)
In RTK, a receiver obtains correction data from a single base station or a local reference network in the same area.
Precise point positioning (PPP)
While accessible from anywhere in the world, receiver initialization for PPP can take up to 30 minutes. Also, a few PPP correction services only provide corrections for satellite clock and orbit errors and not for atmospheric errors, limiting users to a lower accuracy level than with RTK.
Hybrid PPP-RTK
In recent years, new methods have emerged. Hybrid PPP-RTK combines near-RTK accuracy and quick initialization times with the global access of PPP. It relies on a network of reference stations within about 150 kilometers of each other. The stations collect GNSS data and calculate both satellite and atmospheric correction models. The network then broadcasts these corrections via internet, satellites or cellphone towers to subscribers, who can use them to achieve sub-decimeter accuracy.
Each of these methods has advantages and disadvantages (see table 1). RTK, which relies on communication between the user and the local correction service, provides centimeter accuracy over small areas. PPP-RTK and PPP broadcast corrections and require a lighter infrastructure, making them scalable for mass-market and industrial applications. The new services are cheaper and more user-friendly than traditional correction services.
CORS/VRS
Traditional reference networks — often called CORS or virtual reference station (VRS) networks — have long been a source of differential GPS (DGPS) and RTK corrections, mainly for surveying and mapping applications, which requA CORS receiver stands atop the Old River Auxiliary Control Structure, a floodgate system in a branch of the Mississippi in central Louisiana. (Photo: Trimble)ire high accuracies.
“Most CORS in the United States are strictly for providing high-accuracy correction data to GNSS users who need to know their position to less than two centimetres,” said Alex Ngu, applications engineer at Trimble. “However, some — like Utah’s TurnGPS network and the North Carolina Geodetic Survey (NCGS) — have considered dabbling in using them to double up for weather monitoring.” In some regions, such as Japan and Washington state, CORS are also used to study plate tectonics and provide early warning of earthquakes.
CORS receivers often operate in remote locations and may be powered by solar panels. Therefore, they require low power consumption and the ability to configure, run and update remotely. They also need to archive on-board measurements and withstand harsh environments.
Changes in the market
As the market for GNSS corrections changes, so does the role of CORS (continuously operating reference station) networks. They are increasingly used for industrial automation that needs centimetre accuracy, including construction and agriculture. “Now, due to the growth in autonomous systems, such as autonomous cars, people are looking at corrections in a completely different way and with more focus on mass markets,” said Gustavo Lopez, market access manager at Septentrio. Septentrio lets customers choose which correction service to use.
“CORS/VRS networks will keep focus on performance and on adding constellations and signals, but nothing major is expected to happen in these traditional systems,” Lopez said. They will continue to exist because they focus on centimeter-level accuracy for survey, construction, mining, machine control and precision agriculture. “What will really change the market are these new services with 10-cm to 20-cm accuracies, which also offer a new way of delivering the data, namely broadcasting rather than using two-way communication methods.” This helps with adoption by emerging applications, Lopez said.
He predicts that applications needing 10- to 50 centimetre accuracy will migrate to cheaper services, including new consumer applications, advanced driver-assistance systems (ADAS), professional applications such as robotics, UAVs, logistics and internet of things (IoT) applications.
Mobile technologies adopting dual-frequency chipsets also will need correction services. “We will see more and more telecoms interested in providing GNSS corrections as a service, as is already the case in Asia and Europe,” Lopez said. “A few CORS/VRS networks will try to capture part of this emerging applications market by reusing their technology or partnering with other companies to provide a more transparent solution.”
One might think that the rapid expansion of the market for corrections would make it possible for traditional CORS networks to make 1-cm accuracy available at a much lower price. The roadblock is high infrastructure costs, Lopez explained. CORS/VRS networks are expensive to maintain because they require a high density of stations. New services that use broadcasting technology and PPP-RTK positioning modes rely on less dense networks.
New uses for old CORS
VRS – OSR (virtual reference station)
A key benefit of a VRS is that performing RTK positioning across the area it covers does not require guarding a separate GPS base station. Using VRS, the CORS network acts essentially as a continuous reference station within the entire network, enabling RTK positioning using a single rover in the field.
Randy Osborne, VRS network manager at Louisiana State University’s Center for GeoInformatics, reports a growth in new applications beyond surveyors. VRS expanded to precision agriculture, and then into applications such as lidar and UAVs. “We are also seeing strange applications that we never thought of. For example, plumbing companies use it to navigate underground from a truck that has a position on the network, and then they vector from the truck underground into pipelines,” Osborne said. Subscribers also include companies performing survey work for fracking and petrochemical projects.
OSR vs. SSR
Most GNSS correction services are based either on the observation state representation (OSR) or on a state space representation (SSR) of the errors. OSR and SSR use different techniques, delivery mechanisms and core technologies.
OSR (observation state representation)
Legacy GNSS correction service providers supply OSR correction services; examples are RTK and networked RTK (RTN). They rely on transferring corrected GNSS observations from the nearest reference station to the rover using a standardized format. They focus on a geographic region and target surveying, machine control and precision agriculture, providing centimetre level accuracy up to about 30 kilometres of the nearest reference station. Because these services require bi-directional communications and large bandwidth, it is hard to ramp them up for mass-market applications.
SSR (space state representation)
By contrast, new players in the market for correction services, as well as some of the larger legacy ones, provide SSR correction services. SSR uses a network of reference stations to model major errors over large areas. They then transfer this model to the rovers, which create local error models and apply them to their GNSS observations. Depending on the service, accuracy ranges from less than 5 cm to 20 cm, convergence times from 10 seconds to 30 minutes, and coverage from continental to global. Because SSR corrections are broadcast, they can be more easily distributed through internet connections and L-band satellite channels. Because all the rovers rely on the same stream of GNSS correction data, SSR services work well for mass-market applications. The growth in SSR technology is driven mainly by the needs of the automotive industry but is sufficiently generic for adoption in other markets.
The challenge of vertical accuracy
While OSR and SSR have comparable accuracies on a horizontal plane, they differ greatly in their vertical accuracy and initialization times, Osborne said. “When we look at CORS for active control and positioning in the National Spatial Reference System, we are mainly trying to get a handle on the vertical part, as it is the hard problem to solve,” he said.
High-precision vertical accuracy is a challenge for any GNSS-based method. Conventional surveying is still the gold standard. With differential levelling, like with digital levels, results in millimetres are possible. Post-processed GNSS, using data from a good geometry of CORS or base data, can yield results under 2 cm vertical, as can real-time OSR methods like RTK and RTN. SSR solutions, like PPP and hybrids, are presently achieving 5 cm at best. An Achilles heel for SSR vertical solutions is the lack of data for localized sources of error, like tropospheric conditions. Semi-dense networks of CORS can feed ionospheric data to speed PPP convergence, but not the level of tropospheric data needed to match the vertical results that OSR and conventional methods can.
As NEW algorithms and computer modelling techniques emerge the quality and precision of positioning will improve. The technology has been slow to change and as Cloud computing costs continue to fall, real time computing on a worldwide basis. New correction and modelling and data utilization platforms will continue to emerge enabling better use of data, better auditing systems and lower cost of access will make positioning faster and easier at all levels…..survey, static control and autonomous vehicles.