Improving Accuracy In Satellite Navigation Systems

Increasing dependency on the global navigation satellite system (GNSS) constellations is raising concerns about what happens when signals are unavailable, even for short periods of time.

GNSS systems affect our daily lives in ways we often don’t see, from location services to cell phone timing. In fact, these satellites have become a necessary part of critical infrastructure, and higher accuracy is now making its way into mass-market applications. But increased dependency on GNSS also makes us vulnerable to any situation – intentional or not – if connections are interrupted.

“Most people are probably familiar with the use of GNSS for getting from point A to point B,” said Greg Wolff, senior product line manager, frequency and time systems at Microchip Technology. “But when you move into areas like aviation and railway and maritime, these critical infrastructure sectors are incredibly dependent on GNSS-delivered services.”

Improvements to both signal structure and terrestrial equipment are helping to improve accuracy and redundancy. Recent developments have focused both on improving positioning accuracy and timing availability for wider consumer and industrial deployment.

GNSS’s scope: PNT
GNSS systems are typically known by their everyday regional names like GPS, Galileo, and GLONASS. In addition to assisting with navigation, they’re also a critical source of timing for a number of infrastructural applications, and the overall domain of applications is sometimes referred to as positioning, navigation, and timing, or PNT.

As a navigation tool, our everyday use continues as it has been. Accuracy has been somewhat limited, but we’ve been able to backstop GPS with inertial measurement units (IMUs) – accelerometers, gyroscopes, and magnetometers – as well as map-matching, which helps to ensure that the positioning decisions comport with what maps say is possible.

Agriculture, where maps don’t help, has been one of the first industries to adopt GNSS on a more precise basis with self-driving farm equipment largely guided by satellite. But the range of applications accessible by GNSS could grow with better accuracy.

The timing side of things, meanwhile, is critical for a number of applications. It’s long been essential for cellular systems, but that will increase with 5G. “With 5G, the number of points that need accurate timing is growing tremendously,” said Wolff.

5G is important for other infrastructure, as well. “When I say critical infrastructure, I’m referring to a 5G-network operator, airport operations, railway, maritime data centers, and now, very, very big government networks,” said Wolff. “The Department of Homeland Security has gotten very involved in making sure we’re doing the right things with our critical infrastructure to make sure that we’re receiving GNSS signals in a secure manner.”

There are two ongoing challenges that, left unaddressed, make these critical applications vulnerable. One is the availability of the GNSS signal itself. The signal from the satellite is relatively weak and is easily blocked by buildings and other items interfering with visibility into the sky. In addition, the signals can be jammed either by local interference or, more insidiously, by intentional jamming. And they can be spoofed, substituting incorrect data for the true signal.

The other challenge is accuracy. That picture is made somewhat more complicated because it’s possible for accuracy to be intentionally crippled by the military. “The U.S. military defense department is able to switch GPS to accurate mode or to an inaccurate mode,” said Botho Graf zu Eulenburg, former CEO at Sapcorda (now owned by U-Blox.) “This means that GPS could be several hundreds of meters off.”

Intentional inaccuracy aside, limitations arise nowadays more from the environment. The ionosphere appears to get the lion’s share of blame for distortions because it’s an electrically active environment that affects the arrival time of the signals.

“Charged particles in the ionosphere disturb and delay GNSS signals, largely contributing to positional errors,” said Gianmarco Zanda, IoT product manager at Telit.

The troposphere also interferes with signals, in particular due to weather – most of which occurs within that layer. “In the troposphere, we have moisture and things that will also have an impact on the signal path and the signal delay,” said zu Eulenburg.

Fig. 1: Layers in the atmosphere. The ionosphere is not a distinct layer, spanning various areas in the thermosphere and mesophere.

Variations in the satellite orbits also can contribute to inaccuracy. “Together with its GPS information, [the GNSS system] broadcasts the precise orbit of the satellite,” zu Eulenburg continued. “But unfortunately, it’s updated only once or twice a day. If it is half a meter off, then the accuracy on the ground is half a meter off, as well.”

Clock inaccuracies are a further source of error. “Once the satellite clock is off one nanosecond, then it is 30 cm off, as well, in the distance measurement,” said zu Eulenburg. “It is not as stable as engineers would wish. The typical measurement is only about 10 meters accurate.”

Various approaches to improving accuracy have resulted in differing service levels. Aside from the deliberate inaccurate mode, there is the level of accuracy provided by cellphones, which aided by maps can be in the range of 2 to 15 meters.

There also have been a series of systems over the years to enhance accuracy. But getting to that level takes a fair bit of assistance from accompanying technologies, and several improvements have made this possible.

Maintaining time
While IMUs and maps help to backup GNSS signals when they’re not available, those systems don’t help infrastructure that’s looking to GNSS for timing. Different backups are required to ensure continuity of timing. One way of backing up so-called “live-sky” (or “open-sky”) timing is through terrestrial timing. That means sending the same timing information through conventional earth-bound networks.

“There’s a trend not only to make the cell phone equipment itself that’s receiving the GPS signal more resilient, but to reduce the number of GPS receivers in the network,” said Wolff. “This is possible because there are now ways to transport time over the network extremely accurately, and to do that in a way that can withstand network disruptions.”

The ability to do so has improved to the extent that network timing could be even more reliable than GNSS timing. “Do you just trust that the time coming from those constellations is good?” asked Wolff. “The answer that countries are coming to is, ‘No, we’re going to construct a tiny network that can compare multiple timing sources.’ Maybe instead of making the GPS receiver the primary source of time at that particular location, the GPS receiver becomes the second backup, or becomes even the third source of time.”

That creates a problem because network latency will affect the reported time. To address this, the timing packets are time-stamped so the delays encountered in the various network hops can be calculated and removed from the actual time when the report arrives.

Another alternative source of timing can come from local atomic clocks. These may be based on different elements, with differences in cost and accuracy. Such systems can effectively act as flywheels if the “true” timing source disappears for a while.

“The cesium [atomic] clock is giving us frequency, but it’s not telling me what the time is,” said Wolff. “So we can marry the clock and GPS signal together. And if the GPS signal goes away, we can continue to generate time, with the stability of the frequency coming from the cesium clock.”

The use of such systems isn’t completely automated, however. Individuals act as timing managers, using tools from companies like Microchip to monitor timing sources and to make sure that alternatives are engaged when there is an issue with any of the timing sources.

Fig. 2: Management tools allow GNSS signals to be backed up by terrestrial timing information to maintain timing in the event of GNSS disruptions. Source: Microchip

Multiple GNSS bands
With respect to improving position accuracy, there are at least two approaches in play. The first is simply to have multiple signals at different frequencies transmitted from the satellite. This is a more modern approach, and so not all satellites are capable of it.

By using more than one frequency, multi-path and interference issues can be factored out of the calculations more easily, thereby tightening up accuracy. “Use of dual frequency devices can remove all major ionospheric errors as described in the User-Equivalent Ranging Error (UERE) calculation (i.e. from 6 to 3 meters for single frequency, to 0.08 to 0.03 meters for dual frequency),” said Zanda. “Since the ionospheric error is among the major error sources in ‘open sky’ conditions, the use of two frequencies greatly improves the overall accuracy of the module.”

While the different regional GNSS systems operate on similar frequencies, although not identical ones, they are all within the so-called L-band. The use of the L1 band is well-established, but combining it with the L5 band gives an overall more accurate result.

“The L5 band—and equivalent global counterparts—has improved the signal structure and made it more robust when compared to the L1 band, making it less prone to degradation and multipath errors,” said Zanda. “When reflections are caused by buildings, the different structure of the L5 signals is especially useful because the multiple bounced signals do not usually ‘overlap,’ and the receiver is able to discriminate the correct one.”

The L5 band also provides a safety net as well as protection from other radio signals. “The redundancy of multiple frequencies offers a backup in case one band fails and also helps the receiver to identify deprecated signals or accidental faults,” noted Zanda. “Another major advantage of L5 usage is that it’s broadcast inside the band reserved for aviation safety services and is thus protected from radio interference.”

Fig. 3: The different GNSS frequency bands as used by GPS, Galileo, and GLONASS. Source: U-Blox

Terrestrial accuracy assistance
Beyond that, there is a system whereby corrections can be generated and transmitted. This is done through the use of reference stations around the globe whose position is known to millimeter levels. While such systems have been in place for some time, they continue to evolve, trading off deployment cost and positioning accuracy. An earlier example of these was the real-time kinetics (RTK) system.

“RTK was something that was invented beginning of the 2000s, and this is what is largely known in the surveying world,” said Cornelia Waldecker, product manager, services at U-Blox. “A few years later, a technology called ‘PPP’ for ‘precise point positioning’ came in. A big difference between those two technologies is that RTK networks have a fairly short distance between the reference stations, which is roughly 50 kilometers. When you look at the classical PPP service, the distance distances between the reference stations is multiple thousands of kilometers, which means that the PPP network cannot very well estimate atmospheric conditions simply due to the fact that the distances between the reference stations are so far.”

The latest version of this is called RTK-PPP, and it splits the difference between the two, with stations hundreds of kilometers apart. The accuracy is better than PPP, although less than RTK. But this is intended as a mass-market service, and, as Waldecker explained, “For most applications in a mass market, which are very dynamic, you don’t need the last centimeter. But you need consistent accuracy throughout larger regions, like a whole continent.”

By monitoring the GNSS signals at those stations, errors can be calculated and corrections can be broadcast. “We spread so-called reference stations equally over a wide region,” explained zu Eulenburg. “And these reference stations are GPS receivers – very good ones with very good antennas. And 24 hours a day, they receive all the GNSS satellites that are flying by and send those observation data to a data center.”

Additional “integrity” stations are used to monitor all of the reference stations to make sure that nothing appears suspicious with any of the stations or satellites. “We’re using those stations totally independently from our system to check whether the results are correct,” said zu Eulenburg. “As soon as we discover any kind of misbehavior, we will send to the car, ‘Be careful, this cannot be used.’”

Those corrections have been broadcast for a time over the cellular system using a dedicated protocol. But now they are also being broadcast from geostationary satellites in non-GNSS constellations. Even so, they have to be updated frequently.

“The satellites fly 20,000 km from the earth with a speed of 20,000 km per hour, and the atmosphere is very dynamic,” noted zu Eulenburg. “So we need to deliver new correction data every five seconds.”

Access to the corrections is by subscription. The streams are encrypted, meaning that a subscription key is required to receive the services. The corrections aren’t offered directly to consumers. Instead, they are used by GPS service providers whose customers receive the corrected information.

The keys for unlocking the stream have to be global, because this is a broadcast signal. That raises the challenge of rescinding access when a subscription lapses. This is managed by changing the keys periodically.

“The keys are delivered over a special MQTT [a transport protocol] topic that the device needs to subscribe to in order to receive the very latest ones,” explained Franco de Lorenzo, principal product manager at U-Blox. “The keys are dynamic and rotating every four weeks, so if access is blocked, the keys are valid only up to the validity time.”

Fig. 4: The combined system of corrections delivery via both mobile systems and L-band satellites, which are distinct from the GNSS satellites themselves. SPARTN refers to a corrections data format. Source: U-Blox

It’s also possible for vehicles to do some of their own correcting. In theory, cars could even calculate the broadcast corrections themselves, but it would take about 30 minutes to do. As it is, the broadcasts allow a fix in 10 to 15 seconds.

However, there may be local effects that the broadcast corrections cannot account for, such as local weather. Cars are equipped with additional tools to handle that. “There are rain-receiver-aided integrity models in the car to cross-check their observations, and they can experience these local effects even if it’s a thunderstorm 20 km away,” said zu Eulenburg.

And it’s possible for the vehicles to identify readings and corrections that don’t make sense, ensuring the vehicle doesn’t blindly follow incorrect position information. “The receiver in the car will calculate using six or seven GPS satellites coming from different directions,” said zu Eulenburg. “It can cross-check the different satellites.”

Overall, accuracy can go below 10 cm. “We market this with less than 10-cm accuracy, although we know that if we use really high-grade GNSS receivers and antennas in our integrity stations, then we typically achieve between 3 and 6 cm,” said zu Eulenburg. “But in the mass markets, most of the applications use lower-cost antennas and receivers, and there is always some degradation in quality. But we can prove that even with low-cost devices, below 10 cm can be achieved.”

Conclusion
Given this level of accuracy for the mass market, the type of control previously available only to tractors can help regular homeowners, as well. “If you have a very big yard, then you can even program a robotic lawn mower to engrave the weapon of your most beloved football club into the yard,” mused zu Eulenburg.

With increased critical applications relying on GNSS for positioning and for timing, GNSS is becoming more deeply embedded in our infrastructure. Increasing accuracy and robustness will help to keep those systems operating in the face of any GNSS signal disruptions or distortions.