Plasma cold fronts are visible in this April 2001 space storm. The upper left image shows the relatively small number of electrons in the upper atmosphere while the right shows a dramatic increase during the storm.
GNSS and Ionospheric Scintillation
How to Survive the Next Solar Maximum
Figure 1: Scintillation map showing the frequency of disturbances at solar maximum. Scintillation is most intense and most frequent in two bands surrounding the magnetic equator, up to 100 days per year. At poleward latitudes, it is less frequent and it is least frequent at mid-latitude, a few to ten days per year. (Click image to enlarge.)
The approaching solar maximum will produce magnetic storms, ionospheric storms, and disruptions to radio signals that directly affect the world’s technical infrastructure, including GNSS. The authors propose a method for evaluating the ability of GPS receivers to operate through scintillation before the next solar maximum arrives in 2013.
The sun has its own seasons, and the stormy season will soon be upon us. Every 11 years, the sun enters a period of increased activity called the solar maximum.
During this period, the far ultraviolet (FUV) portion of the solar spectrum intensifies, making our ionosphere denser and thicker. Frequent solar flares eject up to 10 billion tons of plasma at speeds approaching 1,000 miles per second. Flare-generated, high-energy protons and x-rays reach the earth nearly instantly.
Flare-generated, high-energy electrons will produce intense broadband bursts of radio waves from HF to above the L-band. Called the sunspot cycle, this period of activity is the result of a solar dynamo in which electric currents and magnetic fields are built up in the outer layer of the sun and then destroyed in energetic outbursts.
The next sunspot maximum is currently predicted to arrive in May 2013 and to be a relatively weak maximum in terms of sunspot count — a prediction that would normally trigger sighs of relief. However, many of the most intense solar outbursts have occurred during below-average solar cycles.
The approaching solar maximum will produce magnetic storms, ionospheric storms, and disruptions to radio signals, including the Global Positioning System and other GNSSs, that directly affect our technical infrastructure. In some cases, solar radio bursts will directly interfere with GNSS signals; in other cases, ionospheric and magnetic storms will disrupt radio signals from satellites.
These conditions will fully test for the first time much of the GPS technology installed since the last solar maximum in 2001. During the height of the previous solar cycle, some users of GPS signals were surprised that their receivers were vulnerable — especially the more precise receivers using carrier phase tracking techniques.
For the casual user of GNSS technology in the United States or Europe, who can tolerate an outage of several minutes at most a few times a year, scintillation is not a concern. Other users should be aware that scintillation will affect their receiver operation. (For an example of comparative results from the previous solar cycle, see the article by K. M. Groves et alia in the Additional Resources section near the end of this article.)
In this article we will review the subject of ionospheric scintillation and suggest a method for evaluating GPS receivers before the next solar maximum arrives in 2013.
The number of free electrons is usually expressed as total electron content (TEC), which is the number of free electrons in a rectangular solid with a one-square-meter cross section extending from the receiver to the satellite.
By a quirk of physical fate, the product of the group velocity and phase velocity of the GPS signals is equal to the speed of light squared. So, if the TEC increases, the group velocity slows down and the phase velocity speeds up to keep their product a constant.
A slower group velocity produces ranging errors while a faster phase velocity causes unexpected phase shifts. If the phase shifts are rapid enough, they can challenge the tracking loops in GPS receivers’ phase lock loops. We refer to variations in group delay and phase advance caused by large-scale variations in electron density as signal refraction.
The second effect, signal diffraction, is more complicated. When ionospheric irregularities form at scale lengths of about 400 meters, they begin to scatter GPS signals; so, the radio wave reaches the receiver through multiple paths. The GPS signals on each path will add in a phase-wise sense, causing fluctuations in the signal amplitude and phase.
The same process occurs with light and can be seen in the fuzzy image passing through jet exhaust from a commercial airliner.
Both refractive and diffractive effects are called scintillation. Unfortunately, diffractive scintillation can seriously challenge GPS receivers, causing signal power fades exceeding 30 dB-Hz and fast phase variations.
Who Should Be Concerned?
At high latitudes, the northern lights will disrupt GPS signals. At tropical latitudes, the ionosphere will create its own storms, now made more intense by the denser, thicker ionosphere. Even at mid-latitudes, the ionosphere will experience storms driven by solar flares and magnetic storms.
Storms in the ionosphere present an additional danger to GPS signals when they create irregularities. Fortunately, decades of studying satellite signals in these locations — recently including GPS signals — has left a clear picture of the ionospheric climate. Figure 1 (above right) illustrates where scintillation will most frequently impact GNSS signals.
The greatest danger to satellite signals is at tropical latitudes where ionospheric storms typically form after sunset and last for several hours. During the day, solar heating causes the ionosphere to rise near the equator and then fall under its own weight down magnetic field lines to form two bands of enhanced density on either side of the geomagnetic equator, as shown in Figure 1.
After sunset, an electromagnetic form of the Rayleigh-Taylor instability (upside down water class) forms. The heavy ionosphere supported by horizontal magnetic field lines can suddenly erupt with bubbles, hundreds of kilometers across, that violently surge upward at many hundreds of meters per second, leaving behind an ionosphere filled with irregularities.
This behavior has a seasonal component, being most intense at the equinoxes, but it departs from this pattern in the South American sector where the geomagnetic equator deviates sharply from the geographic equator. The pattern can also be disrupted by magnetic storms, which can generate tropical ionospheric storms after midnight or thrust ionospheric content poleward into mid-latitudes.
Because the ionosphere is the densest and the thickest in two bands surrounding the magnetic equator, as shown in Figure 1, this is where scintillation is most intense.
At high latitudes, the threat to GPS comes during magnetic storms in which blobs of ionosphere from the dayside are swept over the polar cap onto the nightside. During the last solar maximum, magnetic storms were observed to fatten the ionosphere over the dayside United States and then carry blobs of it over the North Pole and polar cap into Europe.
These blobs form irregularities that cause GPS signals to scintillate and pose significant concern for GPS users at high latitudes. In addition to these effects, individual auroral arcs can cause rapid phase variations or even diffractive scintillation.
At mid-latitudes, the threat comes during magnetic storms when sharp ionospheric gradients are formed. These gradients threaten augmentation systems directly and sometimes they form irregularities that cause GPS signals to scintillate.
Unfortunately, we know very little about this threat because during the last solar maximum very few resources were applied to understanding scintillation at mid-latitudes. Despite the low level of ionospheric activity at mid-latitudes implied in Figure 1, one should not assume that no activity exists there.
What Is Scintillation?
. . .
So, what can be done? The first step to designing more robust receivers is to have a simulation that tests how receivers respond to scintillation.
Cornell Scintillation Model
Another obvious approach is to record the entire GPS bandwidth during scintillation and then play back the recordings into receiver tracking loops. The disadvantages of this approach include the possibility that the recordings may not be representative, the fact that the recordings are probably not statistically stationary, and the fact that most receivers do not allow direct access to their tracking loops.
An alternative is to create scintillation from a first-principles phase screen model and apply it to a GPS signal simulator. However, these models are neither well developed nor theoretically mature for strong scintillation.
We have chosen instead to create a statistical model and then to compare the results of the statistical model with empirical data gathered from GPS receivers and from the WIDEBAND satellite project . . .
. . .
Testing in software is efficient for evaluating the validity of our approach but is not particularly useful for testing GPS receivers existing in hardware. To achieve this latter goal, a signal simulator must be employed in a hardware-in-the-loop test, which we will describe next.
Implementation of the Model
We used a GPS signal simulator for the signal source. This simulator allows us to specify the receiver location and dynamics, the satellites present and their orbits, and base signal power. Moreover, the simulator accepts modification to the signal amplitude and phase at 100 hertz, which is sufficient for even fast scintillation. Individual satellites can be controlled with different phase and amplitude time histories.
We first created a history with a unique S4 and τo for each selected satellite and then combined these into a single file. The histories are then passed to another MATLAB function, along with time and pseudorandom noise (PRN) numbers to create a formatted command file called a User Actions File.
The User Actions File is loaded into the simulator where it can be called for a specific scenario. This file automatically creates the changes in signal amplitude and phase.
. . .
Designing Scintillation-Resistant GPS Receivers
If the GPS application does require the carrier phase, we would recommend using a third-order PLL with a pre-detection interval of around 10 milliseconds and a bandwidth of around 10 hertz. These have been shown to be good values for tracking in the presence of scintillations (For test results, see the article by T. Humphreys et alia, 2009b, in the Additional Resources section.)
Another approach is to use designs that remove (wipe off) the navigation data bits. Because the phase change at the bottom of a deep fade approaches a half cycle, a regular squaring-type PLL cannot distinguish between a data bit transition and what might just be a scintillation-induced phase change.
As one might expect, the time between cycle slips can be greatly extended by wiping off the navigation data bits and allowing the PLL to do full-cycle (i.e., non-squaring-type) tracking. In this mode, the PLL knows that abrupt, half-cycle phase changes are noise, not signal.
A practical approach to data bit wiping is to continuously build a database holding the entire 12.5-minute superframe of each observable satellite’s navigation message. Except during the first 20 seconds after an even-hour GPS time crossing (when the satellite ephemerides are refreshed), and during approximately-once-per-day-per-PRN almanac updates, the database should allow prediction of incoming data bits with better than 98 percent accuracy.
The PLL can then be configured to draw on this database only when experiencing phase trauma. In clear space weather, the code can be configured to build up the database.
The new, modernized signals also offer opportunities to design scintillation-robust receivers. For example, the new pilot (i.e., dataless) signals on L2 and L5 contain no data bit transitions; so, no data bit wiping is required. These signals are by design more scintillation-robust than L1 C/A.
One should also avoid the use of dual-frequency receivers that employ codeless, semicodeless, or z-tracking techniques for tracking the L2 signal. The L2 tracking loops of these receivers are particularly vulnerable to scintillation.
As new GPS satellites transmitting the modernized signals are launched, replace the older L1 + L2P(Y) receivers with modern dual-frequency receivers (L1 + L2C or L1 + L5).
Finally, as a minimum, receivers should be designed to determine if they are being affected by scintillation. To do this, fast (50 hertz) amplitude or carrier-to-noise measurements and calculated S4 should be available to the user. Without this capability, users will not be able to diagnose the presence of scintillation in their receiver operation and may confuse scintillation with other problems.
For example carrier phase differential techniques that produce sub-decimeter accuracy are particularly vulnerable. This is especially true in the tropics, but scintillation at GPS frequencies can happen anywhere.
Also, if one is designing or making receivers for applications that depend on truly continuous operation, then a part of the design trade space should be consideration of scintillation. As for users in the market for GPS receivers, if they require truly continuous operations, then knowing how the receivers respond to a scintillating environment becomes a consideration, depending on when and where the receivers will be employed.
In this article we have offered a statistical approach that can be implemented in a hardware-in-the-loop test for evaluating GPS receiver operation in the presence of scintillations. This approach preserves the relationship between amplitude fades and phase fluctuations.
The MATLAB scripts used to develop the amplitude and phase scenarios from the Cornell Scintillation Model can be obtained at <gps.ece.cornell.edu> under the “Space Weather” link. Further information on how to apply the model using a GPS signal simulator can be found in the article by J. Hinks et alia cited in Additional Resources.
As new receivers are designed for modernized GPS signals and other GNSS signals, the Cornell scintillation model needs to be extended. For moderate scintillations, we know that scintillation fades on L1 and L2 are well correlated, with the fades being somewhat deeper on L2 because of its lower frequency.
However, for more intense scintillation we expect the fades at the two frequencies to become more independent. We are currently researching exactly how this happens and modeling the result.
The Cornell Scintillation Model can still be employed to evaluate the tracking capability of single-frequency L1C/A receivers. During the past solar maximum, many users of GPS signals found that their receivers were vulnerable. With the Cornell Scintillation Model, this need not happen again.
For the complete story, including figures, graphs, and images, please download the PDF of the article, above.
 Aarons, J., “50 years of Radio-Scintillation Observations,” Antennas and Propagation Magazine, IEEE, 39(6), 7-12, doi: 10.1109/74.646785, 1997a.
A GSS7700 GPS simulator from Spirent Communications, Paignton, Devon, United Kingdom, was used as the signal source in evaluating the effects of scintillation on hardware receivers.
Copyright © 2016 Gibbons Media & Research LLC, all rights reserved.