During preparation of playback scenarios for the upcoming leap-second event taking place in June, engineers at Racelogic identified a potential pitfall for GNSS engineers. The difficulty arises from the fact that BeiDou uses a different “day number” for the date to apply the leap second, compared with GPS and Galileo. GPS and Galileo use 1-7 as week day numbers, and BeiDou uses 0-6.
If this fact has been missed during development, then the result is that the leap second may be implemented a day early on GNSS engines that are tracking the BeiDou constellation, said Mark Sampson, product manager for Racelogic.
“We tested four different Beidou enabled receivers, from four leading GNSS companies, and none of them appeared to handle the Beidou leap second correctly. This included an engine which originates from China!” Sampson said. “We have since been in contact with two of these companies, who have confirmed that their hardware does have a bug in the leap-second code due to the numbering of the days.”
The error presents itself when the receiver is running on the BeiDou constellation alone, and when the date is June 29 of this year. In some cases, the BeiDou leap second will be adjusted from 2 to 3 seconds from midnight on June 29, which should in fact occur on midnight of June 30. This will result in an error for the reported UTC time of 1 second for the period of this day. In other cases, the leap second was not implemented at all when running on BeiDou alone.
“We have also checked the output of a BeiDou signal generator from a different simulator company, and this too uses the 1-7 range for the BeiDou leap-second date instead of the correct 0-6 range,” Sampson said. “This may explain why a number of commercial receivers appear to have been caught out by this issue.”
Racelogic LabSat3 simulator.
In order to help companies test for this problem, Racelogic has generated simulated RF data for June 29 and 30, starting 15 minutes before midnight. “We have two sets of files. One set contains BeiDou only signals and the other contains a combination of BeiDou and GPS signals,” Sampson said. “Note that on some of the receivers we have tested, when GPS is being tracked as well, the GPS leap-second message overrides the one coming from BeiDou and applies the leap second correctly.”
The scenarios are compatible with Racelogic’s LabSat3 triple constellation simulator, which is available on a free 15-day loan or can be purchased from Racelogic.
Editor’s note: GPS World Innovation editor Richard Langley has co-authored a study, described below, exploring how irregularities in Earth’s upper atmosphere can distort GPS signals, an important step toward mitigation.
The Aurora Borealis viewed by the crew of Expedition 30 on board the International Space Station. The sequence of shots was taken on February 7, 2012 from 09:54:04 to 10:03:59 GMT, on a pass from the North Pacific Ocean, west of Canada, to southwestern Illinois. Image Credit: NASA/JSC
News from the Jet Propulsion Laboratory
When you don’t know how to get to an unfamiliar place, you probably rely on a smartphone or other device with a GPS module for guidance. You may not realize that, especially at high latitudes on our planet, signals traveling between GPS satellites and your device can get distorted in Earth’s upper atmosphere.
Researchers at NASA’s Jet Propulsion Laboratory (JPL), Pasadena, Calif., in collaboration with the University of New Brunswick in Canada, are studying irregularities in the ionosphere, a part of the atmosphere centered about 217 miles (350 kilometers) above the ground that defines the boundary between Earth and space. The ionosphere is a shell of charged particles (electrons and ions), called plasma, that is produced by solar radiation and energetic particle impact.
The new study, published in the journal Geophysical Research Letters, compares turbulence in the auroral region to that at higher latitudes, and gains insights that could have implications for the mitigation of disturbances in the ionosphere. Auroras are spectacular multicolored lights in the sky that mainly occur when energetic particles driven from the magnetosphere, the protective magnetic bubble that surrounds Earth, crash into the ionosphere below it. The auroral zones are narrow oval-shaped bands over high latitudes outside the polar caps, which are regions around Earth’s magnetic poles. This study focused on the atmosphere above the Northern Hemisphere.
“We want to explore the near-Earth plasma and find out how big plasma irregularities need to be to interfere with navigation signals broadcast by GPS,” said Esayas Shume. Shume is a researcher at JPL and the California Institute of Technology in Pasadena, and lead author of the study.
If you think of the ionosphere as a fluid, the irregularities comprise regions of lower density (bubbles) in the neighborhood of high-density ionization areas, creating the effect of clumps of more and less intense ionization. This “froth” can interfere with radio signals including those from GPS and aircraft, particularly at high latitudes.
The size of the irregularities in the plasma gives researchers clues about their cause, which help predict when and where they will occur. More turbulence means a bigger disturbance to radio signals.
“One of the key findings is that there are different kinds of irregularities in the auroral zone compared to the polar cap,” said Anthony Mannucci, supervisor of the ionospheric and atmospheric remote sensing group at JPL. “We found that the effects on radio signals will be different in these two locations.”
The researchers found that abnormalities above the Arctic polar cap are of a smaller scale — about 0.62 to 5 miles (1 to 8 kilometers) — than in the auroral region, where they are 0.62 to 25 miles (1 to 40 kilometers) in diameter.
Why the difference? As Shume explains, the polar cap is connected to solar wind particles and electric fields in interplanetary space. On the other hand, the region of auroras is connected to the energetic particles in Earth’s magnetosphere, in which magnetic field lines close around Earth. These are crucial details that explain the different dynamics of the two regions.
CAScade, Smallsat and IOnospheric Polar Explorer (CASSIOPE) is a made-in-Canada small satellite from the Canadian Space Agency. It is comprised of three working elements that use the first multi-purpose small satellite platform from the Canadian Small Satellite Bus Program. Image Credit: Canadian Space Agency
To look at irregularities in the ionosphere, researchers used data from the Canadian Space Agency satellite Cascade Smallsat and Ionospheric Polar Explorer (CASSIOPE), which launched in September 2013. The satellite covers the entire region of high latitudes, making it a useful tool for exploring the ionosphere.
The data come from one of the instruments on CASSIOPE that looks at GPS signals as they skim the ionosphere. The instrument was conceived by researchers at the University of New Brunswick.
“It’s the first time this kind of imaging has been done from space,” said Attila Komjathy, JPL principal investigator and co-author of the study. “No one has observed these dimensional scales of the ionosphere before.”
The research has numerous applications. For instance, aircraft flying over the North Pole rely on solid communications with the ground; if they lose these signals, they may be required to change their flight paths, Mannucci said. Radio telescopes may also experience distortion from the ionosphere; understanding the effects could lead to more accurate measurements for astronomy.
“It causes a lot of economic impact when these irregularities flare up and get bigger,” he said.
NASA’s Deep Space Network, which tracks and communicates with spacecraft, is affected by the ionosphere. Komjathy and colleagues also work on mitigating and correcting for these distortions for the DSN. They can use GPS to measure the delay in signals caused by the ionosphere and then relay that information to spacecraft navigators who are using the DSN’s tracking data.
“By understanding the magnitude of the interference, spacecraft navigators can subtract the distortion from the ionosphere to get more accurate spacecraft locations,” Mannucci said.
Other authors on the study were Richard B. Langley of the Geodetic Research Laboratory, University of New Brunswick, Fredericton, New Brunswick, Canada; and Olga Verkhoglyadova and Mark D. Butala of JPL. Funding for the research came from NASA’s Science Mission Directorate in Washington. JPL, a division of the California Institute of Technology in Pasadena, manages the Deep Space Network for NASA.
The German Galileo test and development infrastructure GATE has been recertified to serve as a Galileo open‐air test laboratory, for receiver integrity testing (RAIM) for safety‐of‐life (SoL) applications, and for Galileo SIS ICD conformance of signal characteristics and signal quality.
The GATE facility, in Berchtesgaden, is operated by IFEN GmbH. Certification was conducted by TÜV SÜD, an international service corporation focusing on consulting, testing, certification and training.
GATE consists of eight transmitting stations that emit Galileo signals in the GATE test area in Berchtesgaden, as well as two monitoring stations that receive and process these signals.
For application tests, it is essential for GATE to provide constant Galileo specifications for tests, including position accuracy, signal spectrum and navigation data. This is necessary for both test types: tests with the eight “GATE satellites” only and tests with simultaneously usage of the already-existing Galileo satellites in orbit.
The compliance to the specification was verified by the company NavCert GmbH from Braunschweig, Germany, in a recertification of the GATE test bed. Compared to a full certification, taking place every three years, a recertification only verifies the compliance to the specification by the use of random inspections though tests in GATE.
The recertification also includes an audit of the operation processes of the operating company IFEN GmbH. Here, the implementation and adherence to process procedures for GATE operation were verified. This includes questions such as whether a sufficiently technical skilled team is available for operating GATE, if the performed application tests are documented in a reproducible way, and how the GATE team handles non‐conformances to the specification and improvements to the system.
With finalization of the recertification work, the GATE certificate was extended by TÜV SÜD to January 2016. Because of this, GATE customers can rely on the independent verification of the GATE test and development environment for upcoming testing activities, IFEN said.
As an add‐on, customers of IFEN’s NavX‐NCS GNSS simulator benefit from the recertification by obtaining a confirmation from an independent organization (TÜV Süd), reassuring the functionality of GPS and Galileo signal characteristics and signal quality as per SIS ICD, IFEN said.
NASA is soliciting research on remote sensing techniques that use GNSS for studying the Earth’s environment.
Specifically, the announcement says NASA “seeks innovative approaches to the development of Global Navigation Satellite System (GNSS) remote sensing techniques and algorithms to study the Earth’s environment from the ionosphere to Earth’s interior.” The announcement says NASA is seeking to emphasize the use of reflected GNSS signals for the characterization of the Earth’s surface and mitigation of natural hazards.
Notices of Intent are requested by January 20, 2015, and the due date for proposals is March 20, 2015.
NASA solicits research through the release of various research announcements in a wide range of science and technology disciplines. NASA uses a peer review process to evaluate and select research proposals submitted in response to these research announcements. NASA says that researchers can help achieve national research objectives by submitting research proposals and conducting awarded research. Visit the announcement page for details.
The first navigation signal transmission from the fifth Galileo satellite, one of two Full Operational Capability satellites launched into wrong orbits on August 22, was received today.
Stations of the Cooperative Network for GNSS Observation (CONGO) and the International GNSS Service Multi-GNSS Experiment (MGEX) network tracked an E1 signal with a PRN code of E18 this morning. The signal was first tracked at the La Laguna station (LLAG, Tenerife, Canary Islands) at 06:08:00 UTC.
A few moments later, the satellite was also tracked at the Geodetic Observatory Wettzell (WTZ3, Wettzell, Germany) and at the University of New Brunswick (UNBD, Fredericton, Canada). The receivers at all three stations are JAVAD GNSS Triumph receivers.
Analysis of current Galileo satellite visibility at various tracking stations confirms that the active satellite is GSAT0201, also known as Galileo FOC-FM1 or Galileo 5, with COSPAR ID 2014-050A and NORAD ID 40128.
As reported earlier, the perigee of Galileo 5’s orbit was raised in an effort to make the satellite usable for research, at least, and potentially for positioning and navigation.
Topcon Positioning Group said that its latest GNSS reference receiver, the NET-G5, is capable of tracking a new signal from the GLONASS constellation.
The GLONASS-M 55 satellite was launched in June 2014 and is equipped with the experimental payload capable of transmitting signals in the L3 frequency band. Engineers successfully tracked the signal with the NET-G5 receiver during a series of recent tests at the Topcon Technology Center in Moscow. The use of signals in L3 band alongside L1 and L2 bands is expected to further enhance the competitiveness of the GLONASS system.
“Topcon is committed to continually investing in research and development to offer end-users and the industry the most up-to-date solutions,” said Ivan Di Federico, chief strategy officer for Topcon Positioning Systems. “Our premier engineers, scientists and designers bring world’s first products and technologies to market, and the ability of the NET-G5 to track the latest signal — a first for the industry — is an excellent example of that dedication.”
Using Vanguard and Universal Tracking technologies, the NET-G5 receiver incorporates 452-channels capable of tracking the full GNSS signal spectrum, including modernized GPS, GLONASS, Galileo, BeiDou, QZSS and SBAS signals.
The GPS Reflections Group of University of Colorado-Boulder has been awarded the prestigious Prince Sultan Bin Abdulaziz International Creativity Prize for Water. The prize is awarded biannually to acknowledge innovative work that contributes to the sustainable availability of water and the alleviation of the global problem of water scarcity.
The awards will be presented in a ceremony in Riyadh, Saudi Arabia, on December 16, concurrently with the 6th International Conference on Water Resources and Arid Environments (ICWRAE 6), December 16-18, 2014.
Professors Kristine Larson and Eric Small developed a new method to measure water at the Earth’s surface. The research team discovered that standard geodetic GPS instruments are sensitive to hydrological influences. They subsequently developed a cost-effective technique, GPS Interferometric Reflectometry (GPS-IR), to measure soil moisture, snow depth, and vegetation water content around GPS antennas. GPS-IR has the advantage of relying on an existing GPS infrastructure installed by surveyors and geoscientists that covers an increasingly large portion of the global surface.
Larson and Small collaborated with scientists at the University Corporation for Atmospheric Research and the National Atmospheric and Oceanic Administration, also in Boulder.
The team uses the GPS-IR technique to analyze data streams from existing GPS networks in near real-time. Data from hundreds of operational GPS sites are downloaded and processed, yielding estimates of hydrologic variables within 24 hours.
Scientists and government agencies can access this information at the team’s web portal and use the data to improve monitoring and forecasting of hydrologic variables.
Photo: Mark L. Psiaki, Brady W. O’Hanlon, Steven P. Powell, Jahshan A. Bhatti, Todd E. Humphreys, and Andrew Schofield
Spoofing Detection with Two-Antenna Differential Carrier Phase
By Mark L. Psiaki, Brady W. O’Hanlon, Steven P. Powell, Jahshan A. Bhatti, Todd E. Humphreys, and Andrew Schofield
A new method detects spoofing attacks that are resistant to standard RAIM technique and can sense an attack in a fraction of a second without external aiding. The signal-in-space properties used to detect spoofing are the relationships of the signal arrival directions to the vector that points from one antenna to the other. A real-time implementation succeeded against live-signal spoofing attacks aboard a superyacht, the White Rose of Drachs shown above, cruising in international waters.
Concerns about spoofing of open-service GNSS signals inspired early work on simple receiver-autonomous integrity monitoring (RAIM) methods based on the consistency of the navigation solution. Work on new classes of defense techniques began in earnest after the demonstration of a powerful spoofer that is undetectfable by simple pseudorange-based RAIM methods. There has been a sense of urgency to solve the spoofing problem since the Iranians captured a classified U.S. drone in 2011 and made unsubstantiated claims to have spoofed its GPS. Two dramatic field demonstrations of the spoofer developed by author Humphreys and colleagues at the University of Texas, Austin, heightened interest in spoofing detection: one involved deception of a small airborne unmanned autonomous vehicle (UAV), causing it to dive towards the ground; another sent a superyacht off course without raising any alarms on its bridge.
One class of spoofing detection methods uses encrypted signals, their known relationships to the open-service signals, and after-the-fact availability of encryption information. Such techniques require a high-bandwidth communication link between the potential victim of a spoofing attack and a trusted source of after-the-fact encryption information, and may involve significant latency between attack and detection.
Another class of methods uses advanced RAIM-type techniques. Instead of considering only pseudorange consistency, these RAIM techniques examine additional signal characteristics such as absolute power levels, distortion of the PRN code correlation function along the early/late axis, the possible existence of multiple distinct correlation peaks in signal-acquisition-type calculations, and other signal or receiver characteristics. Such methods are relatively simple to implement because they do not require much additional hardware, if any, but some of these strategies can have trouble distinguishing between multipath and spoofing or between jamming and spoofing.
A third class proposes the addition of Navigation Message Authentication bits. These are encrypted parts of the low-rate navigation data message. Such techniques require modification of the navigation data message and can allow long latencies between the onset of a spoofing attack and its detection.
A fourth class exploits the differing signal-in-space geometry of spoofed signals in comparison to true GNSS signals. All spoofed signals typically arrive from the same direction, but true signals arrive from a multiplicity of directions. Some of these methods use receiver antenna motion to achieve direction-of-arrival sensitivity. Others use an array of two or more receiver antennas.
The most powerful of these detection strategies exploit models of the effects on carrier-phase data of antenna motion or antenna-array geometry. This knowledge may be partial because an unknown antenna-array attitude may need to be determined as part of the detection calculation. Their power derives from the high degree of accuracy with which a typical GNSS receiver can measure beat carrier phase.
Goals. This research follows on moving-antenna/carrier-phase-based spoofing detection work. One of our goals has been to remove the necessity for moving parts by using two antennas and processing their carrier-phase data.
A second goal has been to achieve real-time operation. An earlier prototype moving-antenna system (see “GNSS Spoofing Detection,” GPS World, June 2013) used post-processing and completed its spoofing detection calculations days or weeks after the recording of wide-band RF data during live-signal attacks.
A third goal has been to test this system against actual live-signal spoofing attacks to prove its real-time capabilities and evaluate its performance during the two phases of an attack: the initial signal capture and the post-capture drag-off to erroneous position and timing fixes.
Two-Antenna System Architecture
The system consists of two GNSS patch antennas, GPS receiver hardware and software, and spoofing detection signal-processing hardware and software. Figure 1 shows two versions. The left-hand version connects its two patch antennas to an RF switch. The single analog RF output of the switch is input to a GNSS receiver that is standard in all respects, except for two features. First, it controls the RF switch or, at least, has access to the switching times. Second, it employs a specialized phase-locked loop (PLL) that can track the beat carrier phase of a given signal through the phase jumps that occur at the switching times. The right-hand version connects each antenna to an independent GPS receiver, likely connected to a common reference oscillator.
Figure 1. Two configurations:, the RF-switched-signal/single-receiver configuration (left) and the two-receiver configuration (right).
The last element of each system is a spoofing detection signal-processing unit. Its inputs are the single-differenced beat carrier phases of all tracked signals, with differences taken between the two antennas. In the switched antenna system, each difference is deduced by the specialized PLL. In the two-receiver system, the single-differences are calculated explicitly from each receiver’s beat carrier-phase observables.
Except for the final spoofing detection unit, the two-receiver system on the right-hand side of Figure 1 is already available commercially. Typical applications are CDGPS-based attitude/heading determination. Thus, this is the easiest version to implement.
This system could include more than two antennas. A multi-antenna system could have a dedicated RF front-end and a dedicated set of receiver channels for each antenna, as on the right of Figure 1. Alternatively, a multi-antenna system could include an RF switch between any one of the multiple antennas at the command of the receiver. The latter design would entail a slight modification to the specialized PLL to track multiple independent phase jumps for the independent antenna switches.
Principles. The principles used to detect spoofing can be understood by considering and comparing the signal-in-space and antenna geometries shown in Figure 2, the two-antenna system and three GNSS satellites for a typical non-spoofed case, and Figure 3, a spoofed case. The salient difference is that the different GNSS signals arrive from different directions for the non-spoofed case, namely and . They all arrive from the same direction, the direction of the spoofer , for the spoofed case. For detection purposes, the important geometric feature is the projection of each direction of arrival onto the known separation vector between the two antennas, bBA. This projection has a direct effect on the beat carrier-phase difference between the two antennas. In the non-spoofed case, this effect will vary between the different received signals in ways consistent with the attitude of the vector. In the spoofed case, all of these carrier-phase differences will be identical. The spoofing detection algorithm decides between two hypotheses about the carrier-phase differences, one conjecturing a diversity consistent with authentic signals and the other conjecturing the sameness that is characteristic of spoofed signals.
Figure 2. Geometry of two-antenna spoofing detection system and GNSS satellites for non-spoofed case.Figure 3. Spoofed-case geometry of two-antenna spoofing detection system and GNSS spoofer.
Hypothesis Test
The PDF paper on which this article is based presents the non-spoofed and spoofed signal models that form the basis of a hypothesis test, develops optimal estimation algorithms that fit the observed differential beat carrier phases to the two models, and shows how these estimates and their associated fit error costs can be used to develop a sensible spoofing detection hypothesis test. Download the PDF here.
Offline and Live-Signal Testing
We tested a prototype version of the two-antenna system as depicted on the righthand side of Figure 1. The antennas connect to two independent RF front-ends that run off of the same reference oscillator. These RF front-ends provide input to two independent receivers that track each signal using a delay-lock loop (DLL) and a PLL. Figures 4 and 5 show system elements: two GPS patch antennas mounted on a single ground plane with a spacing of 0.14 meters, two RF front-ends — universal software radio peripherals (USRPs) — with a common ovenized crystal oscillator. Digital signal-processing functions are implemented in real-time software radio receivers (SWRX) running in parallel on a Linux laptop, written in C++. Spoofing detection calculations are performed on the same laptop using algorithms encoded in Matlab.
Figure 4. The two antennas of the prototype spoofing detection system mounted on a common ground plane.Figure 5. Signal processing hardware of the prototype spoofing detection system.
A key feature of this architecture is the ability of its real-time software radios’ C++ code to call the spoofing detector’s Matlab tic function and to pass carrier-phase and other relevant data to the tic function. This feature served to shorten the implementation and test cycle for the prototype system by eliminating the need to translate the original Matlab versions of the spoofing detection algorithms into C++. This enabled rapid re-tuning and redesign of the spoofing detection calculations, exploited during the course of live-signal testing.
The Matlab package displays real-time signal authentication information. Figure 6 shows the version of the display used for this study’s culminating live-signal tests. All displays are updated in real time. The upper left, upper right, and lower left plots scroll along their horizontal time axes to keep the most recent 4.5 minutes of data available. The lower right compass updates each time a new spoofing detection calculation is performed. The green dots in the upper left plot indicate that the time between spoofing detections, Δtspf , is nominally 1 second, though sometimes the gap is longer due to lack of a sufficient number of validated single-differenced carrier phases to carry out the calculation. Thus, the nominal update time for all of the plots in this display is 1 second. Faster updates are possible with the Matlab software, but Δtspf was deemed sufficiently fast for this study’s experiments.
The most important panel in Figure 6 is the upper left spoofing detection statistic time history. The magenta plus signs on the plot show the spoofing detection threshold chosen for this case, γth. The computed γ values are plotted as green o’s if they lie above γth and as red asterisks if they lie below. If γ is above γth, the message “GPS Signals Authenticated” is displayed on the plot; if below, the message switches to the spoofing alert: “GPS SPOOFING ATTACK DETECTED!”
Figure 6. Spoofing detector real-time display. Clockwise from top left: the spoofing detection statistic time history γ(t); four diagnostic time histories that include time histories of the number of satellites used for spoofing detection L(t) (blue asterisks), their corresponding GDOP(t) values (magenta o’s), the time increment between spoofing detection tests Δtspf(t) (green dots), and the compass heading ψ(t) as determined from the two-antenna non-spoofed-case solution (black dots); Compass display; and time history of GPS PRN number availability.
The other three panels proved helpful in diagnosing system performance. A low L value (near 4) or a high GDOP value in the upper right panel indicated poorer reliability of the spoofing detection calculations. A correct compass heading in the absence of spoofing provided a check on the system. During spoofing attacks, the compass heading became jumpy, thereby providing another possible indicator of inauthentic signals.
The vertical scale of the lower left panel lists the possible GPS PRN numbers. The presence of a green or red dot at the level corresponding to a given PRN number indicates that one or both receivers is seeing something from that satellite at the corresponding time. If the dot is red, then the returned data are incomplete or are deemed to be insufficiently validated for use in the spoofing detection calculation. If the dot is green, then the data from that PRN have been used in the detection that has been carried out at that time.
Another feature of the prototype spoofing detection system is its ability to record the wide-band RF data from its two antennas. For each spoofing scenario, the raw samples from both USRPs were recorded while the real-time software receiver was performing its signal-processing operations and while the real-time spoofing detector was doing its calculations. These recorded data streams will allow off-line analysis and testing of a re-tuned or completely redesigned spoofing detection system.
Red Team Receiver/Spoofer. The UT Austin spoofer’s attack strategy overlays the spoofed signal on top of the true signals, ramps up the power to capture the receiver tracking loops, and finally drags the pseudorange, beat carrier phase, and carrier Doppler shift off from their true values to spoofed values. Figure 7 shows the pseudorange part of a spoofing attack: cross-correlation of the receiver’s PRN code replica with the total received signal (blue solid curve); the receiver’s early, prompt, and late correlations (red dots); and the spoofer signal (black dash-dotted curve). In the top plot, the spoofer has zero power, and the receiver sees only the true signal. The second and third plots show the spoofer ramping up its power while maintaining its false signal in alignment with the true signal. The spoofer power in the middle/third plot is sufficient to capture control of the three red dots of the receiver’s DLL. In the fourth and fifth plots, the spoofer initiates and continues a pseudorange drag-off, an intentional falsification of the pseudorange as measured by the victim receiver’s DLL.
Figure 7. Receiver/spoofer attack sequence as viewed from a channel’s code offset cross-correlation function. Spoofer signal: black dash-dotted curve; sum of spoofer and true signals: blue solid curve; receiver early, prompt, and late correlation points: red dots.
The spoofer performs drag-off simultaneously on all spoofed channels in a vector spoofing attack that maintains consistency of all spoofed pseudoranges. After the initiation of drag-off, the victim receiver computes a wrong position, a wrong true time, or both, but the residual pseudorange errors in its navigation solution remain small. Therefore, this type of attack is not detectable by traditional pseudorange-based RAIM calculations.
The receiver spoofer hardware consists of a GNSS reception antenna, the receiver spoofer signal-processing unit, and the spoofer transmission antenna (Figure 8).
Figure 8a. Receiver/spoofer hardware: GPS reception antenna on ship’s rear upper deck.Figure 8b. Receiver/spoofer hardware: directional transmission antenna pointed at the ship’s GPS antenna and the detector antenna pair near the defended ship’s antenna. The orientation of the spoofing transmission antenna, combined with its remote location from the receiver/spoofer’s reception antenna, ensured that the spoofer did not self-spoof.Figure 8c. Receiver/spoofer hardware: spoofer electronics, located amidships.
The receiver/spoofer requires tuning of its transmission power levels. If the power is too high, its spoofing attacks will be too obvious. A very high transmitted power could also saturate the front-end electronics of the intended victim, causing it to jam the system rather than spoof it. If transmitted power is too low, it will not capture the victim’s tracking loops, and its spoofing attack will fail. The proper power level depends on the gain patterns of the spoofer transmission antenna and the victim receiver antenna and on their relative geometry.
Attack Test Scenarios. Three sets of tests were conducted to develop and evaluate the spoofing detection system. The first tests started by recording wideband RF GPS L1 data using USRPs. These data were post-processed in two software receivers that recorded the outputs of their signal tracking loops. Afterwards, the Matlab spoofing detection calculations were run using the recorded tracking loop data as inputs. These preliminary tests at Cornell and Austin proved the efficacy of the spoofing detection algorithms. They did not, however, test system performance during the transition from non-spoofed to spoofed signals that takes place at the initiation of a spoofing attack.
The second set of tests was carried out using the first real-time version of the system, after the Matlab spoofing detection calculations were repackaged into a tic function and linked to the C++ real-time software receivers. This set of tests also was unable to probe the system’s performance at the onset of a spoofing attack, before the signal drag-off.
The final set of tests was conducted aboard the White Rose of Drachs in the Mediterranean’s international waters.
The power adjustment tests on June 27 needed a means to decide whether a given attack had captured the tracking loops of the ship’s GPS receiver. The strategy for confirming capture was to perform a noticeable drag-off after the initial attack. We settled on a vertical drag-off as providing the most obvious indication of a successful capture. Successful attacks dragged the receiver’s reported altitude as high as 5,000 meters.
The tests that evaluated spoofer and spoofing detector antenna placements relative to the ship’s GPS antenna were also important to achieving sensible results. Various placements were tried. The most successful relative geometry is depicted in Figure 8.
The placement of the detector antennas relative to the defended antenna is atypical of likely real-world detection scenarios. It is expected that a real-world spoofing detector will be integral with the defended GNSS receiver.
The culminating live-signal attack involved a 50-minute spoofing scenario in which the attacker took the ship — apparently — from the Adriatic to the coast off of Libya. The scenario’s long distance and short duration required a mid-course speed in excess of 900 knots. This spoofing scenario was designed in the simplest possible way, by taking a straight-line course in WGS-84 Cartesian coordinates from the true location to the spoofed location off of Libya. This course took the spoofed yacht position across the Italian and Sicilian land masses and below the Earth’s surface to a maximum depth of more than 23 kilometers.
Obviously, the White Rose was physically unable to execute this maneuver. Its crew would not have needed spoofing detection to realize that its GPS receiver was returning false readings. The main points of this last test were to dramatize the potential errors that can be caused by a spoofer and to check whether the spoofing detector could continue to function under these drastic conditions.
Figure 9 highlights this unusual scenario with two displays from the ship’s bridge, photographed during the attack. The GPS display shows the speed, 621 kn (knots), and the altitude, 7376 m. The chart display shows the yacht on (or rather, below) dry land and halfway across the “insole” of Italy’s boot. It also shows a tremendously long velocity vector, extending beyond the chart.
Figure 9a. The ship’s bridge GPS receiver display during the Libya spoofing scenario.Figure 9b. The GPS-driven chart during the Libya spoofing scenario.
Spoofing Detection Test Results
Various signal output time histories (Figure 10) illustrate the attack sequence and suggest means to evaluate the spoofing detection system. The upper panel plots the fractional portions of the two-antenna spoofing detector’s single-differenced beat carrier-phase time histories, Δϕ1BA, …, ΔϕLBA for the L = 7 tracked PRN numbers 16, 18, 21, 22, 27, 29, and 31. The middle panel plots the amplitude time history of the 100 Hz prompt [I;Q] accumulation vector for PRN 16, as received at Antenna A of the detection system. The bottom panel plots the PRN 16 carrier Doppler shift time history.
Figure 10. Indicators of initial capture and drag-off during Libya spoofing attack, as measured by the spoofing detection receiver.
This was a strong attack in which the spoofer power was 10.7 dB higher than the power of the real signal for PRN 16. The other spoofed signals had power advantages over their corresponding true signals that ranged from 3.3 dB to 13.6 dB, and the spoofer’s mean power advantage was 10.4 dB. Therefore, the onset of the spoofing attack at 196.1 sec is clearly indicated by the sudden jump in (I2+Q2)0.5 on the middle panel. The upper panel shows a corresponding sudden coalescing of the single-differenced beat carrier phases, which implies that the spoofing detection algorithm should have been able to detect this attack.
The spoofer drag-off started at 321.5 sec, as evidenced by the sudden change in the slope of the carrier Doppler shift time history on the lower panel. The period after the initial attack and before the drag-off is delimited by the vertical magenta and cyan dash-dotted lines. During this interval the spoofer waited to capture the receiver’s tracking loops.
The single-differenced phase time histories in the upper plot appear somewhat noisier during the interim pre-drag-off period of the attack than after the start of the drag-off at 321.5 sec. The grey dotted curve for PRN 27 is an exception because it becomes noisy again starting at about 450 sec due to decreased signal power. The increased noisiness of the differential phase time histories during the interim period is probably the result of interference between the true and spoofed signals, which are likely beating slowly against each other. The response of the spoofing detection algorithm during this phase is uncertain because this multipath-like beating between the two signals is not modeled.
Figure 11 demonstrates performance of the spoofing detection algorithm for the Libya attack scenario. The upper panel of the figures is a repeat of the upper panel of the single-differenced beat carrier-phase time histories from Figure 10, except that they are plotted for a longer duration. The lower panel shows the γ(t) spoofing detection statistic time history. It plots the same information that appeared in the upper left panel of Figure 6 during the corresponding real-time detection tests. At 196 sec γ(t) is clearly above the blue dash-dotted spoofing detection threshold γth. At 196.4 sec it is clearly below γth , which indicates a spoofing detection. It remains below γth for the duration of the attack. In this reprocessed version of the detection calculations, γ(t) has been updated at 5 Hz. Therefore, the earliest possible detection point would have been 196.2 sec, which is 0.1 sec after the onset of the attack. This point corresponds to the green dot in the lower panel of Figure 11 that lies slightly above the blue dash-dotted γth line. Theoretically, the system might have detected the attack at this time, but the finite bandwidth of the two receivers’ PLLs caused lags in the transitions of the single-differenced phases in the top plot, which led to the 0.3 sec lag in the detection of the attack. It is encouraging, however, that the spoofing detector worked well during the initial pre-drag-off phase of the attack, from 196.1 to 321.5 sec, despite the added noisiness of the single-differenced carrier phases in the top plot, likely caused by beating between the true and spoofed signals.
Figure 11. Single-differenced carrier-phase time histories (top plot) and corresponding spoofing detection statistic time history (bottom plot) for Libya spoofing attack scenario.
Figure 12 plots the same quantities as in Figure 11, but for a different spoofing attack, a little less overt than the Libya attack. The power advantage of the spoofer ranged from 3.0 to 14.0 dB for the different channels with a mean power advantage = 9.2 dB. It was detected by the system, as evidenced by the convergence of the single-differenced carrier phases at the onset of the attack at 397.5 sec. The spoofing detection statistic in the bottom panel dives near to the γth detection threshold at the onset of the attack and sometimes passes below it, but it does not stay permanently below the threshold until after the time of drag-off, after 531 sec.
Figure 12. Single-differenced carrier phase time histories (top plot) and spoofing detection statistic time history (bottom plot) for a spoofing attack with a slightly lower power advantage than the Libya attack.
The large oscillations of the single-differenced carrier phases during the pre-drag-off initial capture interval from 397.5 to 531 seconds is likely due to beating between the true and spoofed signals. The largest variations occur for PRNs 12 and 31, which are the ones with the lowest spoofer power advantages, 3.2 and 3.0 dB, respectively. Apparently these oscillations cause γ(t) sometimes to take on values slightly above γth during the interval 397.5 sec < t < 531 sec. Thus, the spoofing detector can experience problems in the initial phases of an attack.
Note that the spoofer failed to capture the tracking loops of the ship’s GPS receiver. This is surprising, given the average spoofer power advantage of 9.2 dB above the true signals. We conjecture that the ship’s GPS antenna had lower gain in the low-elevation direction toward the spoofer transmission antenna than did the detector’s antennas. A lower gain would reduce the spoofer power advantage in the ship’s receiver and could explain why the spoofer failed to deceive it.
Many additional spoofing attacks were carried out aboard the ship. The spoofing detector proved finicky. It took quite some time to get the spoofing detection two-antenna system positioned in a sensible place relative to the ship’s GPS antenna so as to be sensitive to nearly the same spoofing signals. In addition, the spoofing detector’s GPS receiver tended to lose lock at the initiation of an attack, prior to signal drag-off. This was likely caused by the large power swings of the received signals due to beating of the true signals against the spoofed signals. This problem went away at higher spoofer power levels. When lock was lost, the software receiver would attempt to re-acquire the signal. Often a reacquisition would succeed only after signal drag-off by the spoofer. Typically, the spoofing detector immediately detected the attack once it had reacquired the spoofed signals that were no longer beating against the true signals due to having been dragged sufficiently far away from them, as in Figure 7. Re-analysis of the recorded data indicated that poor PLL tuning may have caused the losses of lock during the initial attacks. Spoofing detection calculations carried out on the reprocessed data have proved more reliable when implemented with a better PLL tuning.
Two attacks were carried out with only a subset of the visible GPS satellites being spoofed. The first involved spoofing 7 of 9 visible satellites, and the second test spoofed only 4 of 9. The spoofing detection system had trouble maintaining signal lock during the initial part of the first attack. It subsequently reacquired signals and was able to detect the attack successfully after reacquisition. The first attack also succeeded in capturing the ship receiver’s tracking loops as evidenced by spoofing of the yacht to climb off the sea surface. The second attack, with only four spoofed satellites, was not detected by the prototype system, but it succeeded in deceiving the ship’s GPS receiver about its altitude. This latter result indicates a need to modify the detection calculations to allow for the possibility of partial spoofing. In their current form, they assume that all signals are either spoofed or authentic. Of course, in the partial spoofing case it may also be possible to use traditional pseudorange-based RAIM techniques to detect an attack.
Improved detection during pre-drag-off initial phase of attack;
Detection when only a subset of signals are spoofed;
Advanced RAIM techniques;
A real-time prototype of the switched-antenna version;
Detection of a spoofer that uses multiple transmission antennas;
Reacquisition of true signals to recover from a spoofing attack.
Conclusions
A new prototype GNSS spoofing detection system has been developed and tested using live-signal spoofing attacks. The system detects spoofing by using differences in signal direction-of-arrival characteristics between the spoofed and non-spoofed cases as sensed by a pair of GNSS antennas. A spoofing detection statistic has been developed that equals the difference between the optimized values of the negative-log-likelihood cost functions for two data-fitting problems. One problem fits the single-differenced beat carrier phases of multiple received signals to a spoofed model in which the fractional parts of these differences are identical -— in the absence of receiver noise — because the spoofed signals all arrive from the same direction. The other problem fits the single-differenced carrier phases to a non-spoofed model. This second optimal data-fitting problem is closely related to CDGPS attitude determination. The simple difference of the two optimized cost functions equals a large positive number if there is no spoofing, but it equals a negative number if the signals are being spoofed. Monte Carlo analysis of the probability distributions of this difference under the spoofed and non-spoofed assumptions indicates that it provides a powerful spoofing detection test with a low probability of false alarm.
A real-time version of this system has been implemented using USRPs and real-time software radio receivers, and it has been tested against live-signal spoofing attacks aboard a yacht that was cruising around Italy. Successful detections have been achieved in many spoofing attack scenarios, and detections can occur in as little as 0.4 seconds or less. One scenario spoofed the yacht’s GPS receiver into believing that it had veered off of a northwesterly course towards Venice in the Adriatic to a southwesterly course towards the coast of Libya, and at the incredible speed of 900 knots. The spoofing detector, however, warned the crew on the bridge about the attack before the yacht’s spoofed position was 50 meters away from its true position.
The live-signal tests revealed some challenges for this spoofing detection strategy. They occur primarily during the initial attack phase, before the spoofer has dragged the victim receiver to a wrong position or timing fix. If the spoofer power is not much larger than that of the true signals, then beating occurs between the spoofed and true signals during this initial period. This beating can cause difficulties for the receiver tracking loops, making single-differenced carrier phase unavailable. Even when single-differenced phase is available, both the spoofed and non-spoofed models of this quantity can be inadequate for purposes of designing a reliable spoofing detection test.
This article’s new two-antenna spoofing detection system has generated promising real-time results against live-signal spoofing attacks, but further developments are needed to produce a sufficiently reliable detection system for all anticipated attack scenarios. The best defense will likely employ a multi-layered approach that uses the techniques described in this paper along with advanced RAIM techniques that detect additional signal anomalies that are characteristic of spoofing.
Acknowledgments
The authors (brief bios given in online version) thank the owner of the White Rose of Drachs for the loan of his vessel to conduct the live-signal GNSS spoofing detection tests reported here. The crew of the White Rose aided and supported this project in many ways.
Red Team, White Team, Blue Team
Background
Before March 2013, members of the UT Radionavigation Lab and the Cornell GPS Lab didn’t know about gold-plated sinks and spiral staircases at sea. They did know something about spoofing navigation systems and detecting spoofer attacks. The UT group had hacked a helicopter drone at White Sands Missile Range in June 2012, coaxing it to dive towards the ground. The Cornell group had developed a prototype system that could reliably detect all UT Austin attacks, but it was clumsy, having an oscillating antenna and requiring hours of post-processing.
Andrew Schofield, master of the White Rose of Drachs, attended Todd Humphreys’ 2013 South-by-Southwest conference talk on the drone hack and challenged him to go big — bigger than a 1.3-meter drone helicopter. How about a 65-meter superyacht? The result: a summer 2013 Mediterranean cruise that produced intriguing, provocative results.
The UT team had implemented a feedback controller for their spoofer, but they were unable to control the spoofed drone in a smooth, reliable manner. The White Rose cruise offered a chance to test a next level of sophistication: a controlled sequence of lies leading the victim on a precise course selected by the spoofer, different from the one intended by the captain.
The UT team was able to induce inadvertent turns while the ship’s bridge thought it was steering a straight course. They could nudge the yacht onto a wrong course paralleling the desired course. The crew remained unaware of the yacht’s true course because its GPS receiver and GPS-driven charts indicated that she was on her intended route.
The Push for Protection
Andrew Schofield quickly began advocating for a follow-up experiment: a UT Red Team attack against the White Rose GPS and a simultaneous Cornell Blue Team demonstration of real-time spoofing detection.
The Cornell Team, however, faced challenges in transitioning from its initial prototype to a more sophisticated system, one that eliminated the moving parts and that operated in real time. Team members thought they could produce the next system, but had never been quite sure they could make good on their boast.
Development of a second prototype system began with implementation of a new Cornell detection algorithm in Matlab. The first tests of this algorithm involved UT recording and pre-processing of transmissions in an RF chamber that housed the two antennas of Cornell’s second prototype. Cornell applied its new Matlab algorithm to these data and demonstrated off-line spoofing detection.
The remaining hurdle was real-time operation. The original development plan called for translation of the Matlab algorithm to C++ followed by integration with a UT Austin/Cornell real-time software radio. It would be understatement to say that this was an ambitious task for the two-month window that remained until the White Rose cruise.
UT Ph.D. student Jahshan Bhatti steered the team around this hurdle by proposing the direct use of Cornell’s Matlab code in the real-time system. Prior to this, no one had realized that it could be practical to call Matlab from C++ in real time. Mark Psiaki packaged the Matlab spoofing detection software into a single tic function, Jahshan coded the calling C++/Matlab interface, and the team was on track to test spoofing detection in late June 2014.
Spoofer, Detector Clash at Sea
The White Rose would sail from southern France on June 26, setting a course around Italy to Venice. The Cornell Blue Team would have three full days in international waters to demonstrate and evaluate their real-time spoofng detection system. A Ph.D. graduate from UT’s Radionavigation Laboratory would operate the Red Team spoofer, aka the Texas Lying Machine.
In preparation for the voyage, the two teams converged in the White Roses’s home port of Cap-d’Ail. They performed initial shake-down tests of their systems in port. They could not do full live-signal tests in Cap d’Ail because they were still in French territorial waters. Transmission of live spoofing signals in the GPS L1 band is permitted only in international waters, and only if conducted for scientific purposes.
The spoofing and detection tests started in earnest on the morning of June 27 off the southern coast of Italy. The White Rose had passed through the Strait of Messina between Italy and Sicily earlier that day. The initial tests were concerned with antenna geometries and spoofer power levels. Later tests concentrated on serious deception of the White Rose regarding its true course and location.
During the tests, the UT Red team and its spoofer were situated on the White Rose Sun Deck, above and behind the bridge. The Cornell Blue team and its electronics were on the bridge with its two antennas on the roof. A walkie-talkie link between the teams provided coordination of detector operation with spoofing attacks along with feedback about spoofer and detector performance.
Hijacked to Libya!
For the final day of tests, Andrew Schofield suggested sending the spoofed White Rose to Libya as she cruised the Adriatic from Montenegro to Venice — a difference of 600 nautical miles. The target trip time of 50 minutes necessitated a peak speed over 900 knots (1,667 kilometers/hour) after factoring the need to limit initial acceleration and final deceleration; if too large, they might cause the victim receiver’s tracking loops to lose lock and, therefore, the spoofed signals.
The Cornell and UT Austin teams programmed the spoofer for a trip to Libya, and they initiated the attack. The White Rose bridge soon became a scene of excitement. The ship started veering sharply to port, and its velocity vector lengthened until it literally went off the charts. The GPS receiver showed the ship hurrying towards Libya on a collision course with the back of Italy’s boot. The bridge’s GPS receiver displayed speeds that increased through 100 knots, 200 knots, 300 knots — for a yacht with a speed capability of about 15 knots.
The Cornell detector issued a spoofing alert at the onset of the attack, long before the White Rose veered off course. After a few minutes, the detector’s continued successful operation became boring. Of course, boring success is better than exciting failure.
The Cornell system had not been as successful during some of the preceding attacks, and the results from the June voyage suggested avenues for improvement. If new live-signal tests become necessary to evaluate planned improvements, the Red and Blue teams stand ready for a future superyacht cruise.
Mark L. Psiaki is a Professor of Mechanical and Aerospace Engineering. He received a B.A. in Physics and M.A. and Ph.D. degrees in Mechanical and Aerospace Engineering from Princeton University. His research interests are in the areas of GNSS technology and applications, spacecraft attitude and orbit determination, and general estimation, filtering, and detection.
Brady W. O’Hanlon is a graduate student in the School of Electrical and Computer Engineering. He received a B.S. in Electrical and Computer Engineering from Cornell University. His interests are in the areas of GNSS technology and applications, GNSS security, and space weather.
Steven P. Powell is a Senior Engineer with the GPS and Ionospheric Studies Research Group in the Department of Electrical and Computer Engineering at Cornell University. He has M.S. and B.S. degrees in Electrical Engineering from Cornell University. He has been involved with the design, fabrication, testing, and launch activities of many scientific experiments that have flown on high altitude balloons, sounding rockets, and small satellites. He has designed ground-based and space-based custom GPS receiving systems primarily for scientific applications.
Jahshan A. Bhatti is pursuing a Ph.D. in the Department of Aerospace Engineering and Engineering Mechanics at the University of Texas at Austin, where he also received his M.S. and B.S. He is a member of the UT Radionavigation Laboratory. His research interests are in the development of small satellites, software-defined radio applications, space weather, and GNSS security and integrity.
Todd E. Humphreys is an assistant professor in the department of Aerospace Engineering and Engineering Mechanics at the University of Texas at Austin, and Director of the UT Radionavigation Laboratory. He received a B.S. and M.S. in Electrical and Computer Engineering from Utah State University and a Ph.D. in Aerospace Engineering from Cornell University. He specializes in applying optimal estimation and signal processing techniques to problems in radionavigation. His recent focus is on radionavigation robustness and security.
Andrew Schofield is a career Yacht Captain. After completing his degree in Applied Biology and working in the bio-science industry for a year, he left all that behind in 1991 and found a deck hand’s job on a sailing yacht in the Caribbean. Since then he has worked on various yachts in various locations. He has been Captain of the White Rose of Drachs since launch in June 2004. He is President of the Professional Yachting Association, the large yacht professional body, and focuses on the training and certification of crew. In his time at sea GPS has transformed navigation. He feels that the relevance of the work done to detect GPS spoofing cannot be overstated with regard to the safety of life at sea, and he is delighted to have facilitated the voyage during which spoofing detection was proven.
Ambiguity and Environmental Data: Two Further Key Challenges of Multisensor Positioning
By Paul D. Groves, Lei Wang, Debbie Walter, and Ziyi Jiang, University College London
The coming requirements of greater accuracy and reliability in a range of challenging environments for a multitude of mission-critical applications require a multisensor approach and an over-arching methodology that does not yet exist. Part 1 of this article, in the October issue, examined the two key concepts of complexity and context. In this continuation, we complete our overview with exploration of the requirements of ambiguity and environmental data.
Ambiguity occurs when measurements can be interpreted in more than one way, leading to different navigation solutions, only one of which is correct. Any navigation technique can potentially produce ambiguous measurements. The likelihood depends on both the positioning method and the context, both environmental and behavioral. Urban and indoor positioning techniques that do not require dedicated infrastructure are particularly vulnerable to ambiguity. Poor handling of ambiguity results in erroneous navigation solutions and the navigation system can become “lost,” whereby it is unable to recover and may even reject correct measurements.
There are six main causes of ambiguity: feature identification, pattern matching, propagation anomalies, geometry, system reliability, and context ambiguity. Each of these is described in turn below.
Feature Identification Ambiguity. The proximity, ranging, angular positioning, and Doppler positioning methods all use landmarks for positioning. These may be radio, acoustic, or optical signals, or natural or man-made features of the environment. For reliable positioning, these signals or features must be correctly identified.
Digital signals intended for positioning incorporate identification codes. However, where a signal is weak and/or interference is high, it may be possible to use the signal for positioning but not decode the identification information. For signals of opportunity — that is, not designed for positioning — the identification codes may be encrypted, while analog signals do not typically have identifiers. These signals must be identified using their frequencies and an approximate user position, in which case there may be multiple candidates. Even where a signal of opportunity is identifiable, the transmission site may change without warning. For example, Wi-Fi access points are sometimes moved and mobile phone networks are periodically refigured. Thus, there is a risk of false landmark identification.
Environmental features are difficult to identify uniquely. In image-based navigation, man-made features, such as roads, buildings, and signs, are easiest to identify in images due to their line and corner features. However, similar objects are often repeated in relatively close proximity. For example, Figure 18 shows the locations of the five “no entry” signs in a 1,200-meter circuit of Central London streets. Two of the signs are within 20 meters of each other. (Figure numbering continues the sequence beginning in Part 1, October issue.)
Figure 18. “No entry” signs in a 1,200-meter circuit of Central London. (Background image courtesy of Bing maps | Photo: Paul D. Groves, Lei Wang, Debbie Walter and Ziyi Jiang, University College London)
Pattern-Matching Ambiguity. The pattern-matching positioning method maintains a database of measurable parameters that vary with position. Examples include terrain height, magnetic field variations, Wi-Fi signal strengths, and GNSS signal availability information. Values measured at the current unknown user position are compared with predictions from the database over a series of candidate positions. The position solution is then obtained from the highest scoring candidate(s).
An inherent characteristic of pattern matching is that there is sometimes a good match between measurements and predictions at more than one candidate position. Figure 19 and Figure 20 show GNSS shadow-matching scoring maps based on smartphone measurements taken at the same location 40 seconds apart. The scores are obtained by comparing GNSS signal-to-noise measurements with signal availability predictions derived from a 3D city model. In Figure 19, maximum scores (shown in dark red) are only obtained in the correct street, whereas in Figure 20, there is also a high-scoring area in the adjacent street, giving two possible position solutions.
Figure 19. GNSS shadow-matching scoring map – unambiguous case (the cross shows the true position and white areas are indoor locations). (Photo: Paul D. Groves, Lei Wang, Debbie Walter and Ziyi Jiang, University College London)Figure 20. GNSS shadow-matching scoring map – unambiguous case (the cross shows the true position and white areas are indoor locations). (Photo: Paul D. Groves, Lei Wang, Debbie Walter and Ziyi Jiang, University College London)
Figure 21 presents another example, showing the height of a road vehicle derived from a barometric altimeter at three different times. Provided the altimeter is regularly calibrated, it may be used for terrain-referenced navigation (TRN), determining the car’s position along the road by comparing the measured height with a database. However, if only the current height is compared, it will typically match the database at multiple locations within the search area, as the figure shows. The ambiguity can be reduced by comparing a series of measurements from successive epochs, known as a transect, with the database. This approach is applicable to any pattern-matching technique. However, increasing the transect length to reduce the ambiguity also reduces the update rate, and the ambiguity problem can never be eliminated completely.
Figure 21. Height of a car derived from a barometric altimeter at three different times; readings of around 235 m are highlighted. (Photo: Paul D. Groves, Lei Wang, Debbie Walter and Ziyi Jiang, University College London)
Signal Propagation Anomalies. The ranging, angular positioning, and Doppler positioning methods all make the assumption that the signal propagates from the transmitter (or other landmark) to the user in a straight line at constant speed. Significant position errors can therefore arise when these assumptions are not valid due to phenomena such as non-line-of-sight reception, multipath interference, and severe atmospheric refraction. In challenging environments, such as dense urban areas and indoors, multiple signals are typically affected by propagation anomalies, and it is not always easy to determine which signals are contaminated.
Where the position solution is overdetermined (that is, more than the minimum number of signals are received), different combinations of signals will produce different position solutions when there are significant propagation anomalies.
Figures 22 and 23 illustrate this for conventional GNSS positioning using a Leica Viva geodetic receiver, showing the position errors obtained using different combinations of GPS and GLONASS signals. In Figure 22, the receiver is located on a high rooftop and the majority of position solutions are within 15 meters of the mean, with the remainder easily dismissible as outliers. However, in Figure 23, where the receiver is located in a dense urban location, the candidate position solutions are spread over more than 100 meters, and the correct position solution is not clear. The densest cluster of positions is far from both the centroid and the truth. Therefore, anomalous signal propagation may be treated as an ambiguity problem.
Figure 22. GNSS position errors using different combinations of signals in a rooftop environment. (Photo: Paul D. Groves, Lei Wang, Debbie Walter and Ziyi Jiang, University College London)Figure 23. GNSS position errors using different combinations of signals in a dense urban environment. (Photo: Paul D. Groves, Lei Wang, Debbie Walter and Ziyi Jiang, University College London)
Geometric Ambiguity. Geometric ambiguity occurs when more than one position solution may be derived from a set of otherwise unambiguous measurements. Figure 24 shows two examples. On the left, two ranging measurements in two dimensions produce circular lines of position that intersect in two places. On the right, a ranging measurement and a direction-finding measurement are made using the same signal. As direction finding has a 180° ambiguity, the lines of position also intersect at two places.
Figure 24. Geometric ambiguity in two dimensions from two ranging measurements (left), and a ranging and direction-finding measurement (right). (Photo: Paul D. Groves, Lei Wang, Debbie Walter and Ziyi Jiang, University College London)
System Reliability. Navigation subsystems can produce incorrect information for a host of different reasons. Some examples include:
user equipment hardware and software faults;
transmitter hardware and software faults;
out-of-date databases used for pattern matching, including TRN, GNSS shadow matching, and map matching;
wheel slips in odometry;
the effects of passing vehicles and animals on environmental feature visibility, availability and strength of radio signals, and Doppler-based dead reckoning.
Some of these failure modes are easily detectable through the measurements failing basic range checks or being absent altogether. In other cases, faults may be detected by consistency checks within the subsystem. For example, wheel slip may be detected by comparing measurements from different wheels, while Doppler radar and sonar systems typically incorporate a redundant beam to enable the interruption of a beam by a vehicle or animal to be detected.
Subsystems can sometimes output incorrect information that is plausible. An ambiguity thus exists where it is uncertain whether or not a measurement may be trusted. An ambiguity also exists where a fault has been detected, but not its source. Thus, some of the information produced by the subsystem must be incorrect, but some of it may be correct.
Context Ambiguity. As discussed in Part 1 of this article (October issue), the optimum way of processing sensor information depends on the context. However, if context information is used, the navigation solution will then depend on the assumed context. For example, if an indoor environment is assumed, indoor radio positioning and map-matching algorithms that are only capable of producing an indoor position solution may be used. Similarly, if an urban environment is assumed, GNSS shadow matching and outdoor map matching may be selected, resulting in an outdoor position solution. Adoption of pedestrian and vehicle motion constraints can also lead to different navigation solutions.
Context determination is not a completely reliable process. Therefore, to minimize the impact of incorrect context assumptions on the navigation solution, the context should be treated as ambiguous whenever there is significant uncertainty.
Possible Solutions
There is no obvious solution to the ambiguity problem. Instead, different approaches to integrating ambiguous information may be adopted depending on the relative priorities of solution availability, reliability, and processing load. The main approaches, illustrated in Figure 25, are discussed below. They all require the subsystems to present the different measurement hypotheses and their associated probabilities to the integration algorithm.
Figure 25. Methods of handling ambiguous measurements in a navigation integration algorithm. (Photo: Paul D. Groves, Lei Wang, Debbie Walter and Ziyi Jiang, University College London)
Accept or reject the lead hypothesis. The simplest way of handling ambiguous information is to maintain a single-hypothesis navigation solution and consider only the most-probable hypothesis from each subsystem. This is then accepted or rejected based on the following criteria:
Whether the probability of the highest scoring hypothesis above a certain threshold.
Whether the probability of the second-highest scoring hypothesis below a certain threshold.
Whether the highest-scoring measurement hypothesis is consistent with the current integrated navigation solution. (Determinable using measurement innovation filtering.)
Context may be incorporated into this approach by accepting the highest-scoring behavioral and environmental contexts where they meet the above criteria and computing a context-independent navigation solution otherwise.
This approach is processor-efficient, but high integrity and availability cannot be achieved simultaneously. Low acceptance thresholds provide high reliability by rejecting most erroneous measurements, but low solution availability as many good measurements are also rejected. Conversely, high acceptance thresholds provide availability at the expense of reliability.
Accept all hypotheses into a single-hypothesis solution. A probabilistic data association filter (PDAF) accepts multiple measurement or context hypotheses, weighting them according to their probabilities, but represents the navigation solution as the mean and covariance of a uni-modal distribution. The measurement update to the state estimation error covariance matrix accounts for the spread in the hypotheses such that the state uncertainties can sometimes increase following a measurement update.
This approach reconciles the demands of integrity and availability at the price of a moderate increase in processing load. However, the uni-modal navigation solution can sometimes be misleading. For example, if a pattern-matching system determines that the user is equally likely to be in one of two parallel streets, the overall position solution will be midway between those streets.
Multi-hypothesis integration accepting all hypotheses. Multi-hypothesis integration deals with multiple measurement and context hypotheses by spawning multiple integration filters, one for each hypothesis. Each filter is allocated a probability based not only on the probabilities of the measurements input to it, but also on the consistency of those measurements with the prior estimates of that filter. This consistency-based scoring is essential; otherwise the filter hypothesis that inputs the highest-scoring measurement hypotheses will always dominate, regardless of whether those measurements are consistent across subsystems and successive epochs.
A fundamental characteristic of multi-hypothesis filtering is that the number of hypotheses grows exponentially from epoch to epoch. This is clearly impractical, so the number of hypotheses is limited by merging the lowest scoring hypotheses into higher scoring neighbors.
The overall navigation solution is the weighted sum of the constituent filter hypotheses. Each individual filter hypothesis describes a uni-modal distribution. However, the combined navigation solution is multi-modal. Thus, the position probability can be higher in two streets than in the buildings between those streets. This is a clear advantage over the PDAF-based approach, but the processing load is higher.
Multi-modal integration accepting all hypotheses. A multi-modal filter is not constrained to model the states it estimates in terms of a mean and covariance. This enables it to process multiple measurement and/or context hypotheses and represent the result as a weighted sum of the probability distributions arising from the individual hypotheses. Suitable data-fusion algorithms include the Gaussian mixture filter and the particle filter. A key advantage over multi-hypothesis integration is that measurements may be treated as continuous probability distributions instead of as a set of discrete hypotheses. This enables pattern-matching measurements to be integrated more naturally and offers greater flexibility in handling signal propagation anomalies.
A Gaussian mixture filter models the probability distribution of the navigation solution as the weighted sum of a series of multi-variate Gaussian distributions. An example is the iterative Gaussian mixture approximation of the posterior (IGMAP) technique, which has been applied to terrain referenced navigation integrated with inertial navigation.
A particle filter models the probability distribution of the navigation solution using a series of semi-randomly distributed samples, known as particles. Between a thousand and a million particles are typically deployed, with a higher density of particles in higher probability regions of the distribution. Particle filters have been used with a number of different navigation technologies, including TRN, pedestrian map matching, Wi-Fi positioning, and GNSS shadow matching.
Multi-modal integration algorithms offer the greatest flexibility in reconciling the demands of solution availability and reliability, but also potentially impose the highest processing load.
Issues to Resolve
The key challenge in handling ambiguous measurements is determining realistic probabilities for each hypothesis. A probability must also be calculated for the null hypothesis, that is, the hypothesis that every candidate measurement output by the subsystem is wrong. The same applies to ambiguous context.
A feature identification algorithm must allocate a score to every database feature that it compares with the sensor measurements. In practice, only features within a predefined search area, based on the prior position solution and its uncertainty, will be considered. Features scoring above a certain threshold will be possible matches. Similarly, pattern- matching algorithms allocate a score to each candidate position in the search area according to how well the sensor measurements match the database at that point. For correct handling of ambiguous matches, these scores should be as close as possible to the probabilities of the feature match or candidate position being correct.
Feature identification and pattern-matching algorithms can also fail to consider the correct feature or candidate position for several reasons. The correct feature or position may be outside the database search area. It may be absent due to the database being out of date. The sensor may also observe or be affected by a temporary feature that is not in the database, such as a vehicle. The null hypothesis probability must account for all of these possibilities. In practice, it will be higher where there is no good match between the measurements and database.
Signal propagation anomalies affect the error distributions of ranging, angle, and Doppler shift measurements, and the positions and velocities derived from them. These error distributions depend on whether the signals are direct line-of-sight (LOS), non-line-of-sight (NLOS), or multipath- contaminated LOS. However, this is not typically known. Signal strength measurements, environmental context, signal elevation (for GNSS), distance from the transmitter (for terrestrial signals), consistency between different measurements, and 3D city models can all contribute useful information. However, their relationship with the measurement errors is complex, so a semi-empirical approach is needed.
Moving on to reliability, virtually any subsystem can produce false information. The overall probability will typically be very low and thus only significant for high-integrity applications. However, the failure probability will be higher in certain circumstances, in which case the relevant subsystem should report a higher null probability. For example, in odometry, the probability of a wheel slip depends on host vehicle dynamics. Similarly, a radio signal is more likely to be faulty if it is weaker than normal. Repeated measurements, changes to the update interval, and sudden changes in a sensor output are also indicative of potential faults.
Geometric ambiguity is easy to quantify as the candidate solutions have equal probability in the absence of additional information.
As proposed in Part 1, the context determination process should produce multiple context hypotheses, each with an associated probability. Therefore, it is important to ensure that all navigation subsystems that use this context information do so in a probabilistic manner. Thus, where different context hypotheses lead to different values of the measurements output by a navigation subsystem, each measurement hypotheses should be accompanied by a probability derived from the context probabilities.
A further issue to resolve is the relationship between discrete and continuous ambiguity. Ambiguities in feature identification, solution geometry, failures, and context categorization are discrete and are suited to integration filters that treat them as a set of discrete hypotheses. However, the position solution ambiguity in pattern-matching is continuous, that is, the probability density is a continuous function of position, albeit sampled at discrete grid points. This probability distribution may be input directly to a particle filter. However, if the integration algorithm is a uni-modal filter or a bank of uni-modal filters, the probability distribution must be converted to a set of discrete hypotheses. This can be done by fitting a set of Gaussian distributions to the probability distribution. For signal propagation anomalies, their presence or absence is discrete. However, the resulting measurement error distribution is continuous, so a similar approach is appropriate.
The same challenging environments that require multiple navigation subsystems to maximize solution availability, accuracy, and reliability can also induce those subsystems to produce ambiguous measurements. Consequently, the modular integration architecture proposed in Part 1 should be capable of handling ambiguous measurements.
This is discussed further in our IEEE/ION PLANS 2014 paper, “The Four Key Challenges of Advanced Multisensor Navigation and Positioning.”
Environmental Data
Position-fixing systems need information about the environment, sometimes known as a “world model,” to operate. Proximity, ranging, and angular positioning all use landmarks that must be identified. For GNSS and other long-range radio systems, identification codes are determined when the system is designed and incorporated in the user equipment. However, this is not practical for shorter range signals, whether opportunistic or designed for positioning, due to the vast numbers of transmitters available worldwide and the fact that many will be installed during the lifetime of the user equipment. The user equipment will also require information on the characteristics of a signal to enable it to use that signal for ranging. A mobile device equipped with a generic radio or transceiver may be required to download software to enable it to use a proprietary indoor positioning system. For environmental feature-matching techniques, the user equipment requires information to enable it to identify each landmark.
Navigation using landmarks also requires their positions and, for passive ranging, their timing offsets. Signals designed for positioning typically provide this information, but it can take a long time to download (30 seconds for GPS C/A code) and can be difficult to demodulate under poor reception conditions. The positions of opportunistic radio transmitters and environmental features must be determined by other means.
For positioning using the pattern-matching method, a measurement of radio signal strength or a characteristic of the environment, such as the terrain height or magnetic field, is compared with a database to determine position. Therefore, a database providing values of the measured parameter over a regular grid of positions is required. Map matching requires a map database to indicate where the user can and cannot go. GNSS shadow matching requires a 3D city model to predict signal visibility.
Finally, as discussed in Part 1 of this article, mapping is required to determine environmental context information from the position solution and to enable location-dependent context connectivity information (for example, the location of train stations) to be used for context determination.
Possible Solutions
We discuss in turn the environmental data collection and its distribution to the user equipment.
Data Collection. Positioning data may be collected either from a systematic survey or by the users. In either case, regular updates will be required. A systematic survey might be conducted by the subsystem supplier, a national mapping agency, or a private third party. The user will need to pay for the data in some way. It could be included in the equipment cost, via a subscription payment, by accepting advertising, or through general taxation (for some national mapping agency data). For mobile devices, such as smartphones, mapping data may be available for some applications, but not others.
Single-user data collection does not involve user charges, but only provides data for places the user has already visited. A simple approach requires a good position solution to collect mapping data. This can work for applications that normally use GNSS, but require backups for temporary outages. However, it does not work for areas where GNSS reception is poor. Simultaneous localization and mapping (SLAM) techniques can perform mapping without a continuous position solution. However, there are several constraints. First, a good position solution that is independent of the data being mapped is required at some point, usually the start. Second, a navigation system including dead-reckoning technology must be used. Third, locations must be visited repeatedly within a short period of time (to achieve “loop closure”). Finally, only features close to the user can be mapped.
Cooperative mapping by a group of users solves many of the problems of single-user mapping. It can provide individual users with data for places they have not visited before. Distant landmarks can also be mapped more easily by multiple users, particularly where it is necessary to determine a timing offset as well as the location. However, a method for comparing and combining data from multiple users is required.
Data Distribution. For data collected by a systematic survey, there are two main data distribution models: pre-loading and streaming. Pre-loading requires sufficient user equipment data storage to cover the area of operation. New data may have to be loaded prior to a change in operating area, and updates will be required. However, a continuous communications link is not needed.
Streaming requires much less data to be stored by the user and provides up-to-date information, but only where a communications link is available. Although buffering can bridge short outages, navigation data is simply not available for areas without sufficient communications coverage. Continuous streaming can also be expensive. One solution is a cooperative approach using peer-to-peer communications for much of the data distribution. A pair of users traveling in opposite directions along the same route will each have data that is useful to the other. A further possibility is to incorporate local information servers in Wi-Fi access points for exchanging information relevant to the immediate locality. This might be best suited to indoor navigation, where there is an incentive for the building operator to provide the service.
For data collected by a single user, no data distribution is required other than a back-up. For cooperative data collection by multiple users, a method of data exchange is needed. This can be via a central server, communicating either in real time or whenever the user returns to base. It can also be through peer-to-peer communications or through local information servers, where there is an incentive to provide them.
Issues to Resolve
Standardization is a major part of the data management challenge. A multisensor navigation system will typically incorporate multiple subsystems with data requirements. This might include road or building mapping, radio signal information, terrain height, magnetic anomalies, visual landmarks, and building signal-masking information for GNSS shadow matching. There will be a different standard for each type of data. Furthermore, different subsystem suppliers will often use different standards for the same type of data. This is sometimes done for commercial and/or security reasons, so the data may be encrypted. There may also be technical reasons for different data standards. For example, in image-based navigation, different feature recognition algorithms require different descriptive data.
Ideally, all navigation data in a multisensor system should be distributed by the same method. This requires agreement of storage and communication protocols that can handle many different data formats, including encrypted proprietary data and future data formats. Open standards for each type of data should also be agreed, noting that consumer cooperative positioning using peer-to-peer communications and/or local information servers is probably only practical with open data formats. Ideally, the standards should be scalable to enable precisions, spatial resolutions, and search areas to be adapted to the available data storage and communications capacity.
Peer-to-peer data exchange requires a suitable communications link. Bluetooth is the established standard for consumer applications. Classic Bluetooth provides sufficient capacity, but it takes longer to establish a connection than passing pedestrians or vehicles remain within range. Bluetooth low energy can establish a connection quickly, but the data capacity is limited to 100 kbit/s. This is sufficient for some kinds of navigation data, but not others. Professional and military users have more flexibility to select suitable datalinks.
Finally, establishing local information servers requires both standardization and an incentive for the hosts. Demand would be greater if there were applications beyond navigation and positioning. Possibilities include product information in shops and exhibit information in museums, both of which might be provided more efficiently from a local server than the Internet. For home users to provide local information servers, they would also have to benefit from them, a potential “chicken-and-egg” problem. For military applications, local information servers are a potential security risk and a target for attack.
Conclusions and Recommendations
Achieving accurate and reliable navigation in challenging environments without additional infrastructure requires complex multisensor integrated navigation systems. However, implementing them presents four key challenges: complexity, context, ambiguity, and environmental data handling. Each of these problems has been explored and solutions proposed.
Conclusions. In Part 1 of this article, a modular integration architecture was proposed to enable multiple subsystems from different organizations to be integrated without the need for whole system expertise or sharing of intellectual property. Furthermore, context-adaptive navigation was proposed to enable a navigation system to respond to changes in the environment and host vehicle (or user) behavior, deploying the most appropriate algorithms. A new probabilistic approach to context determination was proposed and results presented from a number of context detection experiments.
Here, it has been shown that navigation solution ambiguity can arise from feature identification, pattern matching, propagation anomalies, solution geometry, system reliability issues, and context ambiguity. A number of methods for handling ambiguous measurements in a multisensor navigation system have been reviewed.
Finally, methods of collecting and distributing data such as locations of radio transmitters and other landmarks, information for identifying signals and landmarks, road or building mapping, terrain height, magnetic anomalies, and building signal-masking information (for GNSS shadow matching) have been discussed.
Implementing the ideas proposed in this two-part article requires both standardization and further research. Standardization is needed to enable the communication between modules produced by different suppliers of information such as the integrated navigation solution, sensor measurements and characteristics, calibration parameters, performance requirements, context information, mapping, and signal and feature characteristics.
Further research is needed to support this standardization process, including the identification of a set of fundamental measurement types and their error sources, and the establishment of the best set of context categories for integrated navigation.
Extensive research into context detection and determination is needed, including the measurements to use, the statistical parameters to derive from those measurements, and a set of context association and connectivity rules.
An assessment of the different methods for handling ambiguous measurements is needed, comparing accuracy, reliability, solution availability, and processing load. This will enable the community to determine which methods are suited to different applications.
Finally, there is a need for a practical demonstration of the key concepts proposed in this paper, including modular integration, context adaptivity, ambiguous measurement handling, and collection and distribution of environmental data.
Paul D. Groves is a lecturer at University College London (UCL), where he leads a program of research into robust positioning and navigation. He is an author of more than 60 technical publications, including the book Principles of GNSS, Inertial and Multi-Sensor Integrated Navigation Systems, now in its second edition. He is a Fellow of the Royal Institute of Navigation and holds a doctorate in physics from the University of Oxford.
Lei Wang is a Ph.D. student at UCL. He received a bachelor’s degree in geodesy and geomatics from Wuhan University. He is interested in GNSS-based positioning techniques for urban canyons.
Debbie Walter is a Ph.D. student at UCL. She is interested in navigation techniques not reliant on GNSS, multi-sensor integration, and robust navigation. She has an MSci from Imperial College London in physics and has worked as an IT software testing manager.
Ziyi Jiang was a postdoctoral research associate at UCL until 2014, working on urban GNSS and other projects. He holds a bachelor’s degree in engineering from Harbin University and a Ph.D. in rail positioning from UCL. He now works in finance.
All authors are members of UCL Engineering’s Space Geodesy and Navigation Laboratory (SGNL).
In recent tests, the IDS ran continuously for 24 hours monitoring for potential interference originating from traffic on a nearby highway, SR-522, passing through Woodinville, Washington. “In a single day, the IDS detected two separate instances of a sweep jammer moving along the highway,” noted Loctronix Founder and CEO Michael Mathews. “These discoveries were unexpected, given the relatively short monitoring period and the fact that SR-522 is not a heavy truck-route.”
“The two interference events were likely caused by sweep jammers installed within a vehicle’s interior. The intercepted signals exhibited significant variations in amplitude probably caused by the jammer antenna non-uniform radiation pattern as the jammer vehicle passed through the IDS antenna beam pattern,” Mathews added.
Intentional interference is designed to prevent a GPS receiver from acquiring and tracking signals. The use of jammers is in the U.S. is illegal; however, they can still be purchased for as little as $30. Thousands of GPS jammers are purportedly in use throughout Europe and several parties have been caught illegally jamming GPS in the U.S.
Loctronix developed the IDS to identify, characterize, and ultimately geolocate GPS interference. When interference is present, the system analyzes the interference for signal structure and notifies operators if the threat is significant.
The IDS is highly portable, simple to use, and cost-effective, Loctronix said. The system is based upon the Loctronix ASR-2300 software defined radio platform, making it readily configurable (from a single mobile detector to a multi-sensor network array) to monitor additional GNSS bands and, potentially, cellular bands.
In the video below, Michael B. Mathews, Ph.D., CEO and founder of Loctronix, tells GPS World about the IDS at the ION GNSS+ Conference, held September in Tampa, Florida.
Complexity and Context: Key Challenges of Multisensor Positioning
By Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London
The next generation of navigation and positioning systems must provide greater accuracy and reliability in a range of challenging environments to meet the needs of a variety of mission-critical applications. No single navigation technology is robust enough to meet these requirements on its own, so a multisensor solution is required. Four key challenges must be met: complexity, context, ambiguity, and environmental data handling.
Although many new navigation and positioning methods have been developed in recent years to address GNSS shortcomings in terms of signal penetration and interference vulnerability, little has been done to bring them together into a robust, reliable, and cost-effective integrated system.
New positioning techniques investigated over the past 15 years include:Wi-Fi; ultra-wideband; phone signals; television and other signals of opportunity; Bluetooth; lasers, and dead reckoning; pedestrian dead reckoning (PDR) using step detection; pedestrian and activity-based map matching; magnetic anomaly matching; and GNSS shadow matching.
There have also been improvements to existing technologies: visual navigation, dead-reckoning algorithms, micro-electro-mechanical systems, inertial sensing with cold-atom technology, nuclear magnetic resonance gyros, distance-measuring equipment, Loran, Doppler with Iridium, multiple GNSS constellations, network assistance, and augmentation by commercial pseudolite systems.
In the next generation, a universal navigation system might be expected to provide position within 3 meters at any location with a very high reliability. No single positioning technology is capable of meeting the most demanding application requirements. Radio signals may or may not be subject to obstruction, attenuation, reflection, jamming, and/or interference. Known environmental features, such as signs, buildings, terrain height variation, and magnetic anomalies, may or may not be available for positioning. The system could be stationary, carried by a pedestrian, or on any type of land, sea, or air vehicle. Furthermore, for many applications, the environment and host behavior are subject to change. A multisensor solution is thus required.
A robust, reliable, and cost-effective integrated system must meet four key challenges:
Complexity. How to find the necessary expertise to integrate a diverse range of technologies, how to combine technologies from different organizations that wish to protect their intellectual property, how to incorporate new technologies and methods without having to redesign the whole system, and how to share development effort over a range of different applications.
Context. How to ensure that the navigation system configuration is optimized for the operating environment and host vehicle (or pedestrian) behavior when both are subject to change.
Ambiguity. How to handle multiple hypotheses, including measurements of non-unique environmental features, pattern-matching fixes where the measurements match the database at multiple locations, and uncertain signal properties, such as whether reception is direct or non-line-of-sight (NLOS).
Environmental Data Handling. How to gather, distribute, and store the information needed to identify signals and environmental features and define their points of origin or spatial variation.
Complexity
Achieving robust positioning in challenging environments potentially requires a large number of subsystems. For example, Figure 1 shows the possible components of a pedestrian navigation system using sensors found in a typical smartphone. Figure 2 shows possible components of a car navigation system using equipment already common on cars and other suitable low-cost sensors. Some technologies are common to the two platforms, while others differ.
Figure 1. Potential components of a pedestrian navigation system using smartphone sensors. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)Figure 2. Potential components of a car navigation system using commonly available equipment and other low-cost sensors. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)
Any multisensor navigation or positioning system needs integration algorithms to obtain the best overall position solution from the constituent subsystems. These algorithms must not only input and combine measurements from a wide range of subsystems, but also calibrate systematic errors in those subsystems. Designing the integration algorithms therefore requires expertise in all of the subsystems, which can be difficult to establish in a single organization. The more subsystems there are, the more of a problem this is.
The expert knowledge problem is compounded by the fact that different modules in an integrated navigation system are often supplied by different organizations, who may be reluctant to share necessary design information if this is considered to be intellectual property that must be protected. In a typical smartphone, one company supplies the GNSS chip, another supplies the Wi-Fi positioning service, a third organization supplies the mapping, the network operator provides the phone-signal positioning, a fifth company provides the inertial and magnetic sensors, and a sixth company produces the operating system. Because of lack of cooperation between these different organizations, useful information gets lost. For example, GNSS pseudo-range measurements are not normally available to application developers.
A further issue is reconfigurability. To minimize development costs, manufacturers share algorithms and software across different products, incorporating different subsystems. They also want to minimize the cost of adding new sensors to a product to improve performance. Similarly, researchers want to compare different combinations of subsystems. However, with a conventional system architecture, modifications must be made throughout the integration algorithm each time a subsystem is added, removed, or replaced. The more subsystems there are, the more complex this task becomes.
For a given application, different subsystems may also be used at different times. For example, a smartphone may use Wi-Fi positioning indoors and GNSS outdoors and may deploy different motion constraints and map matching algorithms, depending on whether the device is carried by a pedestrian or traveling in a car. Different integration algorithms for different configurations are more processor efficient, but also require more development effort. Conversely, an all-subsystem integration algorithm is quicker to develop, but can waste processing resources handling inactive subsystems.
Modular Integration. The solution to these problems is a modular integration architecture, consisting of a universal integration filter module and a set of configuration modules, one for each subsystem. The integration filter module would be designed by data fusion experts without the need for detailed knowledge of the subsystems. It would accept a number of generic measurement types, such as position fixes and pseudo-ranges, with associated metadata. The configuration modules would be developed by the subsystem suppliers and would convert the subsystem measurements into a format understood by the filter module and supply the metadata. They would also mediate the feedback of information from the integration filter to the subsystems. The metadata comprises the additional information required to integrate the measurements such as
the measurement type and any coordinate frame(s) used.
a sensor identification number (to distinguish measurements of the same type from different sensors).
statistical properties of the random and systematic measurement errors.
identification numbers and locations of transmitters and other landmarks.
A key advantage of this approach is that subsystems may be changed without the need to modify the integration filter. Provided the new subsystem is compatible, all that is needed is the corresponding configuration module.
Figure 3 shows an example of a modular integration architecture for a combination of conventional GNSS positioning, GNSS shadow matching, Wi-Fi positioning, and PDR. As well as providing measurements and associated statistical data to the integration filter module, the configuration modules feedback relevant information to the subsystems. Shadow matching works by comparing measured and predicted signal availability over a number of candidate positions, so requires a search area to be specified using other positioning technologies. PDR uses information from other sensors, where available, to calibrate the coefficients of its step length estimation model and correct for heading drift. Conventional GNSS positioning can also benefit from position and velocity aiding to support acquisition and tracking of weak signals in indoor and urban environments.
Figure 3. Modular integration of conventional GNSS, shadow matching, PDR, and Wi-Fi positioning for pedestrian navigation (different colors denote potentially different suppliers). (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)
In principle, each subsystem configuration module could simply supply a position fix to the integration filter module with an associated error covariance. However, other forms of measurement generally give better results. For conventional GNSS positioning, the advantages of tightly coupled (range- domain) integration over loosely coupled (position-domain) are well known.
PDR is a dead-reckoning technique, so measures distance traveled rather than position. Consequently, providing measurements of position displacement and direction can avoid cumulative errors in the measurement stream.
GNSS shadow matching and some types of Wi-Fi positioning use the pattern-matching positioning method. This scores an array of candidate position solutions according to the match between the measured and predicted signal availability or signal strength. Although the output of these algorithms is in the position domain, a likelihood distribution can provide more information for the integration filter than a simple mean and covariance.
Other navigation and positioning techniques generate further types of measurement, including velocity, attitude, specific force, angular rate, range rate, and bearings and elevations of features. The types of measurement depend on the positioning method.
A universal integration filter must operate without prior knowledge of which measurements it must process and which states it must estimate. Consequently, it must reconfigure its measurement vector, state vector, and associated matrices according to the measurements available, using the metadata supplied by the configuration module. This capability is sometimes called “plug and play,” and a number of prototypes have been developed by different research groups.
The integration filter must be capable of implementing either error-state or total-state integration, depending on the measurements available. In error-state integration, one of the subsystems, such as inertial navigation, provides a reference navigation solution. The integration filter estimates corrections to that solution using the measurements from other subsystems. In total-state integration, the integration filter estimates the position and velocity directly, and an additional configuration module provides information on the host vehicle (or pedestrian) dynamics.
Modular integration algorithms could form part of a wider modular integrated navigation concept in which subsystem hardware and software is shared across a range of applications.
Issues to Resolve
A critical requirement for the successful implementation of modular integration is an open-standard interface for communication between the universal filter and configuration modules. This enables modules produced by different organizations to work together. To realize the full benefits of modular integration, in terms of interoperability and software re-use, there should be a single standard covering the consumer, professional, research, and military user communities and spanning all of the application domains air, sea, land, indoor, underwater, and so forth. A standard developed by one group in isolation is unlikely to meet the needs of the whole navigation and positioning community, while the development of multiple competing standards defeats the main purpose of modular integration.
This interface should be defined in terms of fundamental measurement types, such as position, velocity, and the ranges, bearings, and elevations of signals and features. However, there are many different coordinate systems that may be used and positioning may be in 2 or 3 dimensions, while ranging measurements may be true ranges or pseudoranges. Ranging and angular positioning measurements may be differenced across transmitters or landmarks, differenced across receivers or sensors, or double differenced across both.
A universal interface must support every measurement type that requires different processing by the filter module. However, it need not support formats that are easily convertible. Thus, there is no need to support both the north, east, down, and east, north, up conventions. There are two main approaches to defining the fundamental measurement types:
A minimal number of very generic measurement types with metadata used to describe how these should be processed by the integration filter.
A large number of more specific measurement types for which the processing methodology is already known.
For each measurement type, an error specification must be defined. For error sources assumed to be white, a standard deviation or power spectral density (PSD) is required. For correlated errors, such as biases, information on the time correlation is required alongside variances and covariance information. The interface standard should include every conceivable error source. Unused errors can simply be zeroed. The filter module should then use the error specification to determine which error sources to model and how.
Obtaining reliable navigation sensor error specifications can be difficult. Manufacturers often provide only limited information, while performance in the field can be different from that in the laboratory due to vibration and electromagnetic interference. For new positioning techniques, the error behavior may not be fully understood, while complex error behavior can be difficult to measure. Adaptive estimation techniques provide only a partial solution. Even where the error behavior is well known, it can be too complex to practically model within the estimation algorithm. This could represent a fifth challenge.
For subsystems used as the reference in an error-state integration filter, such as an inertial navigation system (INS), the errors will typically be correlated across the different components of the subsystem navigation solution, for example position, velocity, and attitude. Furthermore, to represent the error behavior within an integration algorithm, it is necessary to model the error properties of the underlying sensors, accelerometers and gyroscopes in the case of inertial navigation. Thus, it is likely that additional compound measurement types for reference system data will be needed.
For pseudorange measurements, an issue to consider is the synchronization of different transmitter and receiver clocks. Clocks in receivers for different types of signal, such as GNSS and Loran, may or may not be synchronized with each other. Also, the transmitter clocks are typically synchronized in groups. For example, the GPS satellite clocks are synchronized with each other, as are the GLONASS satellite clocks, but GLONASS is not currently synchronized with GPS. For optimal integration of pseudoranges from different sources, this information must be conveyed to the integration filter.
The interface standard for communication between the filter and configuration modules must also support feedback of information from the integration filter to the subsystems, via the configuration modules. The integrated position, velocity, and attitude solution, with its associated error covariance, is useful for aiding many different subsystems. Therefore, a generic standard for this should be defined. Conversely, the feedback to the subsystems of calibration parameters estimated by the integration algorithm is sensor specific, so should be incorporated in the definitions of the fundamental measurement types.
The user requirements, such as accuracy, integrity, continuity, solution availability, update rate, and power consumption, can vary greatly between applications. For example, accuracy is important for surveying, integrity for civil aviation, solution availability for many military applications, and power consumption for many consumer applications. This impacts the design of the whole navigation system. Different modules could be used for different applications. However, it is more efficient if the components adapt to different environments. Figure 4 shows how requirements information can be disseminated in a modular integrated navigation system.
Figure 4. Modular integration architecture incorporating requirements. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)
An open-standard interface specification should be able to handle any conceivable navigation and positioning system. However, it is more efficient if the components adapt to different environments. Similarly, there will be differences in the error magnitudes that an integration filter can handle and in its capability to handle non-Gaussian error distributions. Variations in fault detection and integrity monitoring capability can also be expected. Consequently, there must be a capability specification for each filter module and a protocol for handling mismatches between the measurements and the filter module, and a means to certify that a filter module actually has the claimed capabilities. (Further discussion of modular integration may be found in our IEEE/ION PLANS 2014 paper, “The Four Key Challenges of Advanced Multisensor Navigation and Positioning,” and the Journal of Navigation paper, “The Complexity Problem in Future Multisensor Navigation and Positioning Systems: A Modular Solution.”)
Context
Context is the environment that a navigation system operates in and the behavior of its host vehicle or user. Examples include a pedestrian walking (behavior) in an urban street (environment), a car driving at highway speeds on an open road, and an airliner flying high above an ocean.
Context is critical to the operation of a navigation or positioning system. The environment affects the types of signals available. For example, GNSS reception is poor indoors while Wi-Fi is not widely available outside towns and cities. In underwater environments, most radio signals cannot propagate so acoustic signals are used instead. Processing techniques can also be context dependent. For example, in open environments, non-line-of-sight (NLOS) reception of GNSS signals or multipath interference may be detected using consistency checking techniques based on sequential elimination. However, in dense urban areas, more sophisticated algorithms are required and may be enhanced using 3D city models. GNSS shadow matching only works in outdoor urban environments.
Navigation using environmental feature matching is inherently context-dependent as different types of feature are available in different environments. Suitable algorithms, databases, and sensors must be selected. For example, terrain referenced navigation (TRN) uses radar or laser scanning in the air, sonar or echo sounding at sea, and barometric pressure on land. Map matching requires different approaches for cars, trains, and pedestrians. Similarly, algorithms and databases for image-based navigation depend on the types of feature available, which vary with the environment.
Behavioral context is also important and can contribute additional information to the navigation solution. For example, cars normally remain on the road, effectively removing one dimension from the position solution. Their wheels also impose constraints on the way they can move, reducing the number of inertial sensors required to measure their motion. Similarly, PDR using step detection depends inherently on the characteristics of human walking. Using PDR for vehicle navigation or vehicle motion constraints for pedestrian navigation will produce errors.
Host vehicle behavior is also important for tuning the dynamic model within a total-state navigation filter and for detecting faults through discrepancies between measured and expected behavior. Within a GNSS receiver, the behavior can be used to set tracking loop bandwidths and coherent correlator accumulation intervals, and to predict the temporal variation of multipath errors. The antenna placement on a vehicle or person can also affect performance.
Historically, context was implicit; a navigation system was designed to be used in a particular type of vehicle, handling its associated behavior and environments. However, many navigation systems now need to operate in a variety of different contexts. For example, a smartphone moves between indoor and outdoor environments and can be stationary, on a pedestrian, or in a vehicle. Similarly, a small surveillance drone may operate from above, amongst buildings, or even indoors. At the same time, most of the new positioning techniques developed to enable navigation in challenging environments, are context-dependent. To make use of these techniques in practical applications (as opposed to research demonstrators), it is necessary to know the context.
Context-Adaptive Navigation
The solution to the problem of using context-dependent navigation techniques in variable-context applications is context-adaptive navigation. As shown in Figure 5, the navigation system detects the current environmental and behavioral context and, in real time, reconfigures its algorithms accordingly. For example, different radio positioning signals and techniques may be selected, inertial sensor data may be processed in different ways, different map-matching algorithms may be selected, and the tuning of the integration algorithms may be varied.
Figure 5. A context-adaptive navigation system. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)
Previous work on context-adaptive navigation and positioning focused on individual subsystems and concerned either behavioral or environmental context, not both.
For example, there has been substantial research into classifying pedestrian motion using inertial sensors to enable PDR algorithms using step detection to estimate the distance travelled from the detected motion. The context information may also be used for non-navigation purposes.
Typically, orientation-independent signals are generated from the accelerometer and gyro outputs. Statistics such as the mean, standard deviation, root mean squared (RMS), inter-quartile range, mean absolute deviation, maximum−minimum, maximum magnitude, number of zero crossings, and number of mean crossings are then determined from a few seconds of data. Frequency-domain statistics may also be used. Finally, a pattern recognition algorithm is used to match these parameters to the stored characteristics of different combinations of activity types and sensor locations.
Detection of road-induced vibration using accelerometers has been used to determine whether or not a land vehicle is stationary, while a calibrated yaw-axis gyro can be used to determine when a vehicle is travelling in a straight line. Indoor and outdoor environments may be distinguished using GNSS carrier-power-to-noise-density ratio (C/N0 ) measurements. Wi-Fi signals might also be used for environmental context detection.
Context Detection Experiments
We have conducted a number of different context-detection experiments using GNSS, Wi-Fi, and accelerometers. Full details are presented in our ION GNSS+ 2013 paper, “Context Detection, Categorization and Connectivity for Advanced Adaptive Integrated Navigation,” and in our PLANS 2014 paper. Here, some highlights from the results are presented.
GNSS. GNSS data was collected at five locations inside and immediately outside UCL’s Grant Museum of Zoology; these are shown in Figure 6. C/N0 measurement data was collected from all GPS and GLONASS signals received by a Samsung Galaxy S3 Android smartphone. About 60 seconds of data was collected at each site. Figure 7 presents histograms of the C/N0 measurements and Table 1 lists the means and standard deviations.
Figure 6. Locations for the GNSS indoor/outdoor context detection experiment. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)Figure 7. GNSS C/N0 measurement distributions at sites inside and immediately outside UCL’s Grant Museum of Zoology. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)Table 1. Means and standard deviations of GNSS C/N0 measurements inside and outside UCL’s Grant Museum of Zoology. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)
As expected, the average received C/N0 is lower indoors than outdoors and lower deep indoors than near the entrance. Furthermore, the standard deviation of the C/N0 measurements is larger outdoors than indoors and also larger near the entrance to the building than deep indoors. Thus, both the mean and the standard deviation of the measured C/N0 across all GNSS satellites tracked are useful both for detecting indoor and outdoor contexts and for distinguishing between different types of indoor environment.
Indoor/Outdoor Detection, Wi-Fi. Tests in and around several UCL buildings have shown no clear relationship between Wi-Fi SNRs and environmental context. However, as the environment changes, there is a rapid change in the Wi-Fi SNRs over a few epochs. For a user moving from inside to outside of a particular building, those signals which originate inside go from strong to weak, while many of those from neighboring buildings become stronger. Consequently, Wi-Fi signals could potentially be used to detect context changes instead of the absolute context. This is useful for improving the overall robustness of context determination.
To test this, Wi-Fi data was collected using a Samsung Galaxy S3 smartphone along a route with both indoor and outdoor sections and a context-change score calculated from the last six epochs of data at 1-second intervals.
Context-change score results are presented in Figure 8. The large blue blocks indicate when the user was outside and the smaller blue block shows when the user was in the building’s basement, a very different Wi-Fi environment. As can be seen, there are clear peaks in the “context change” score whenever the user moves between indoor and outdoor contexts.
However, there are also peaks when the user enters and leaves the basement, so the technique is sensitive to false positives and must be combined with other context detection techniques to be used reliably.
Figure 8. Context-change score computer from Wi-Fi SNR measurements. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)
Behavioral Detection, Accelerometers. The use of accelerometers to detect behavioral context is well established. However, by looking at the vibration spectra, more information can be extracted. For these experiments, specific force data was collected using an Xsens MTi-G IMU/GNSS device, the mean subtracted to remove most of the gravity, and a discrete Fourier transform obtained using the MATLAB function fft. Figures 9 and 10 respectively show the vibration spectra of the specific force magnitude for an IMU on a table and held by a stationary pedestrian. The table spectrum is approximately white, whereas the pedestrian data shows peaks between 6 and 10 Hz.
Figure 9. IMU spectra on a table. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)Figure 10. IMU spectra, stationary pedestrian. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)
Figures 11 and 12 respectively show the vibration spectra of a stationary Vauxhall Insignia car, and a stationary urban electric train. Here, the individual accelerometer spectra are shown. In each case, the x-axis was pointing forward, the y-axis to the right and the z-axis down. The car exhibits a lot of vibration at frequencies above 10 Hz due to its engine, whereas the dominant train vibration peak is around 1.5 Hz, with smaller peaks at 15 Hz, 25 Hz, 33 Hz, and 50 Hz, the mains power frequency. Thus, the two vehicles are very different from each other and also from the pedestrian. Figure 13 then shows the vibration spectrum of the car moving on a high-speed road. As might be expected, there is much more vibration when moving with broad peaks below 15 Hz due to road vibration and above 15 Hz due to engine vibration.
Figure 11. Specific force frequency spectrum of a stationary car. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)Figure 12. Specific force frequency spectrum of a stationary train. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)Figure 13. Specific force frequency spectrum of a car traveling on a high- speed road. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)
Finally,Figure 14 shows the vibration spectra on an escalator at an underground rail station. The IMU was in the trouser pocket of a pedestrian. Vibration at a range of frequencies below 30 Hz can be seen and it was observed that the resonant frequencies vary between individual escalators.
Figure 14. Specific force frequency spectrum on an escalator. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)
Issues to Resolve
Despite the work done with individual sensors, a multisensor integrated navigation system that adapts to both environmental and behavioral context remains at the concept stage. Realizing this in a practical system requires both effective context determination and a set of context categories standardized across the whole navigation and positioning community.
The first step in the standardization process is to establish a framework suitable for navigation and positioning. Each context category must map to a configuration of the navigation system; otherwise, it serves no purpose. Multiple categories may map to the same configuration as different navigation systems will respond to different context information. In an autonomous context-adaptive navigation system, the context categories must also be distinguishable from each other.
Figure 15 shows the relationships in a five-attribute framework, comprising environment class, environment type, behavior class, vehicle type, and activity type. The environmental and behavioral contexts are treated separately because they perform fundamentally different roles in navigation. Environmental context concerns the availability of signals and other features that may be used for determining position whereas behavioral context is concerned with motion.
Figure 15. Proposed attributes of a context category. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)
Context may be considered at different levels. Sometimes it is sufficient to consider broad classes such as indoor or aircraft. In other cases, more detail is needed, specifying the type of indoor environment or the type of aircraft. Therefore, a two-level categorization framework, comprising class and type is proposed. The behavioral context comprises the vehicle type and the activity undertaken by that vehicle. A common set of classes containing separate vehicle and activity types is thus proposed. For pedestrian navigation, different parts of the body move quite differently, so the sensor location on the body is analogous to the vehicle type.
The broad classes of environmental and behavioral context are relatively obvious. We therefore propose that the community adopts the classes in Table 2. Standardization at the type level requires further research to determine:
which context categories a navigation system needs to distinguish between in order to optimally configure itself;
which context categories may be distinguished reliably by context detection and determination algorithms.
Table 2. Proposed environment and behavior classes. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)
Effective Context Determination. The reliability of current context detection techniques is typically 90−99%, with some context categories easier to detect than others. For the purposes of controlling a navigation system, this is relatively poor. Furthermore, context detection research projects have typically considered a much smaller range of context categories than a practical context-adaptive navigation system would need. Generally, the more categories there are, the harder it is to distinguish between them.
To make context determination reliable enough for context- adaptive navigation to be practical, a new approach is needed. Firstly, the context should be detected using as much information as possible, maximizing both the range of sensors used and the number of parameters derived from each sensor.
Environmental context detection experiments have largely focused on GNSS and Wi-Fi signals. Other types of radio signal; environmental features detected using cameras, laser scanners, radar, or sonar; ambient light; sounds; odors; magnetic anomalies, and air pressure could all be used. Context may also be inferred by comparing the position solution with a map, provided both are sufficiently accurate.
Behavioral context detection experiments have generally used inertial sensors. As shown earlier, this could be taken further by analyzing different frequency bands and, where possible, separating the forward, transverse, and vertical components. Other motion sensing techniques, such as visual odometry and wheel-speed odometry could be used. Context information, such as vehicle type, can also be determined from the velocity, attitude, and acceleration solutions.
Considering every combination of environment type, vehicle type (or pedestrian sensor location), and activity type produces potentially tens of thousands of different context categories — too many to practically distinguish using context detection techniques alone. However, the number of context categories that must be considered may be reduced substantially by using association, scope, and connectivity information, making the determination process much more reliable.
Association is the connection between the different attributes of context. Certain activities are associated with certain vehicle types and certain behaviors are associated with certain environments; an airliner flies, while a train does not, and flying takes place in the air, not at the bottom of the sea.
For a particular application, the scope defines each context category to be required, unsupported, or forbidden. This enables forbidden context categories to be eliminated from the context determination process and required categories to be treated as more likely than unsupported categories.
Connectivity describes the relationship between context categories. If a direct transition between two categories can occur, they are connected. Otherwise, they are not. Thus, stationary vehicle behavior is connected to pedestrian behavior, whereas moving vehicle behavior is not because a vehicle must normally stop to enable a person to get in or out. Context connectivity is directly analogous to the road link connectivity used in map matching and a similar mathematical formulation may be used. In practice, it is best to represent the connectivity as continuously valued transition probabilities rather than in Boolean terms. This facilitates recovery from incorrect context determination and enables rare transitions between context categories to be represented.
Location-dependent connectivity takes the concept a stage further by considering that many transitions between context categories happen at specific places. For example, people normally board and leave trains at stations and fixed-wing aircraft typically require an airstrip to take off and land. Thus context transition probabilities may be modeled as functions of the position solution, provided the positioning and mapping error distributions are adequately modeled and the probability of transitions occurring at unusual locations is considered.
Finally, for maximum robustness, the whole context determination process should be probabilistic, not discrete. The system should maintain a list of possible context category hypotheses, each with an associated probability. Multiple context detection algorithms should be used, each based on different sensor information. The detection algorithms should also output multiple context category hypotheses with associated probabilities. The context determination algorithm should then produce a new list of context category hypotheses and their probabilities by combining:
the previous list of hypotheses and their probabilities;
the hypotheses and probabilities output by the context detection algorithms;
context association, scope, and connectivity information.
Figure 16 illustrates the concept. When there is insufficient information to determine a clear context category, the list of context hypotheses and their probabilities will be output to the navigation algorithms. The handling of ambiguous information in navigation systems is discussed in Part 2.
Figure 16. Probabilistic context determination. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)
Context Adaptivity and Integration
The practical implementation of a complex multisensor navigation system for a multi-context application requires context-adaptive navigation to be incorporated into a modular multisensor integration architecture as described earlier. To enable different modules to adapt to changes in context, the architecture shown in Figure 4 should be extended to supply context information to the configuration modules, integration filter, and dynamic model from the system control module, alongside the user requirements. The configuration modules can then pass the context information onto the subsystems where necessary. Standardization of context categories and definitions across the navigation and positioning community is essential for this. Distribution of context information is useful even for single-context applications as it enables suppliers to provide modules that are optimized for multiple contexts.
The modular integration architecture must also support the context detection and determination process, allowing all subsystems to contribute. The configuration modules should therefore provide context detection information to a context determination module, as shown in Figure 17. The scope information should be supplied by the system control module.
Figure 17. Context-adaptive modular multisensor integration architecture. (Photo: Paul D. Groves, Lei Wang, Debbie Walter, Henry Martin, and Kimon Voutsis, University College London)
Potential architectures for this are discussed in our PLANS 2014 paper.
Ambiguity and Environmental Data
Part 2 of this article, appearing in the November issue, explores the two remaining key challenges and forms conclusions and recommendations.
Paul Groves is a lecturer at University College London (UCL), where he leads a program of research into robust positioning and navigation. He is an author of more than 50 technical publications, including the book Principles of GNSS, Inertial and Multi-Sensor Integrated Navigation Systems, now in its second edition. He is a Fellow of the Royal Institute of Navigation and holds a doctorate in physics from the University of Oxford.
Lei Wang is a Ph.D. student at UCL. He received a bachelor’s degree in geodesy and geomatics from Wuhan University. He is interested in GNSS-based positioning techniques for urban canyons.
Debbie Walter is a Ph.D. student at UCL. She is interested in navigation techniques not reliant on GNSS, multi-sensor integration and robust navigation. She has an MSci from Imperial College London in physics and has worked as an IT software testing manager.
Henry Martin is a Ph.D. student at UCL. His project is concerned with improving navigation performance from a low-cost MEMS IMU. He is interested in inertial navigation, IMU error modelling, multi-sensor integration and calibration algorithms. He holds a master of mathematics degree from Trinity College at the University of Oxford and an MSc in advanced mechanical engineering from Cranfield University.
Kimon Voutsis is a Ph.D. student at UCL. He is interested in pedestrian routing models, human biomechanics, and positioning sensor performance under high accelerations, particularly IMUs and GNSS. He holds an MSc in geographic information science (UCL). His Ph.D. project investigates the effects of pedestrian motion on positioning.
All authors are members of UCL Engineering’s Space Geodesy and Navigation Laboratory (SGNL).
A Look at High-Latitude and Equatorial Ionospheric Disturbances of GPS Signals
By Yu Jiao, Yu (Jade) Morton, Steve Taylor, and Wouter Pelgrum
INNOVATION INSIGHTS by Richard Langley
THE EARTH’S IONOSPHERE. It’s both a blessing and a curse. Together with the magnetosphere, it helps to protect life on our planet from the damaging outpour of particle and electromagnetic radiation from the sun. In particular, it absorbs a lot of the extreme-ultraviolet (EUV) radiation arriving at the Earth. In fact, that is primarily how the ionosphere is formed. The EUV energy strips off the outer electrons of atmospheric gases producing a plasma of free electrons and ions.
The ionosphere has another beneficial role in that it permits long distance radio communication using high-frequency (HF) or shortwave signals. Although its use is in decline since the advent of the Internet, HF is still in use by some broadcasters and military organizations and is indispensible during natural disasters when electricity grids and network links go down.
But the ionosphere can be a pain, too, particularly for GNSS users. The signals from GNSS satellites must travel though the ionosphere on their way to receivers on or near the Earth’s surface. The signals are perturbed by the presence of the free electrons causing an advance in the phase of a signal’s carrier and a delay in the arrival of the pseudorandom noise code modulation (due to the refractive index being frequency dependent or dispersive) and so there is a contribution to carrier-phase and pseudorange (code) measurements, which must be accounted for when determining positions, velocities, and time (PVT) from the measurements.
Again, since the ionosphere is a dispersive medium, by linearly combining simultaneous measurements (either pseudoranges or carrier phases) on two frequencies such as the GPS L1 and L2 frequencies, an observable virtually free of ionospheric effects can be constructed and used for PVT determinations. This approach does require, however, a dual- or multi-frequency receiver. Single-frequency receivers (or the post-processing of single-frequency data) require the use of a model to account for the ionospheric biases as much as possible. The GPS navigation message, for example, includes values of the parameters of a simple ionospheric model. But, on average, its accuracy is only around 50%. More accurate ionospheric corrections can be acquired from elsewhere, even in real time, such as those from satellite-based augmentation systems.
But there is another ionospheric effect that can play havoc with GNSS signals: scintillations. These are rapid fluctuations in the amplitude and phase of the signals caused by small-scale irregularities in the ionosphere. When sufficiently strong, scintillations can result in the strength of a received signal dropping below the threshold required for acquisition and tracking or in causing problems for the receiver’s phase lock loop resulting in many cycle slips.
The occurrence of scintillations depends on many factors including solar and geomagnetic activity, time of year, time of day, and geographical location. In particular, scintillations are most prevalent in equatorial and polar (Arctic and Antarctic) regions. And the processes involved are not fully understood, hindering our ability to model and predict scintillations.
In an effort to help improve the monitoring, mapping, and modeling of scintillations, a team of researchers led by Prof. Jade Morton is monitoring high-latitude and equatorial scintillations and they discuss some of their preliminary results in this month’s column.
“Innovation” is a regular feature that discusses advances in GPS technology and its applications as well as the fundamentals of GPS positioning. The column is coordinated by Richard Langley of the Department of Geodesy and Geomatics Engineering, University of New Brunswick. He welcomes comments and topic ideas. Write to him at lang @ unb.ca.
Among other effects of the Earth’s ionosphere on GPS and other GNSS signals, scintillation is potentially the most problematic. Ionospheric scintillation refers to the random amplitude and phase fluctuations of radio signals after propagating through plasma irregularities. These irregularities occur more frequently in high-latitude and equatorial regions, especially during solar maxima. Occurrence of scintillation is difficult to predict and model because of the complexity of the ionosphere’s internal mechanisms and solar activities that are the driving forces of space weather phenomena. GNSS signals are particularly vulnerable to scintillation, as strong scintillation can severely impact the acquisition and tracking processes in GNSS receivers, causing degradation in positioning accuracy and even loss-of-lock. With the increasing reliance on GNSS applications, understanding the characteristics of ionospheric scintillation and its effects on GNSS signals and receivers has become an important topic and has gained worldwide attention from both ionospheric scientists and GNSS engineers.
Since 2009, our research group has established several ionospheric scintillation monitoring and data collection systems located in high-latitude and equatorial regions. The results presented here are based on data collected from a specialized commercial dual-frequency GPS ionospheric monitoring receiver at Gakona, Alaska (62.4°N, 145.2°W), and a commercial multi-system, multi-frequency GNSS ionospheric monitoring receiver located at Jicamarca, Peru (11.9°S, 76.9°W).
Measurements are filtered to remove slowly varying trends caused by satellite-receiver dynamics, receiver oscillator errors, the background ionosphere and troposphere gradient, and other potential contributions from multipath and man-made interferences. Scintillation events above preset threshold levels from the filter outputs are extracted for analysis. The threshold levels are set based on two commonly used scintillation indices, the S4 index and σφ , which are defined as the standard deviations of the detrended signal amplitude and carrier phase to represent the magnitude of signal intensity and phase fluctuation, respectively. In the study discussed in this article, the thresholds for S4 and σφ are 0.15 and 15°, respectively for high-latitude measurements. For low-latitude data, the threshold for S4 is raised to 0.2 to accommodate stronger amplitude scintillation, while the threshold for σφ remains 15°.
From data collected at Gakona, between August 2010 and March 2013, we extracted 655 amplitude and 2,355 phase-scintillation events from 657 equivalent days of data, while from data collected at Jicamarca, we extracted about 830 amplitude and 1,100 phase-scintillation events from 190 days of data collected from November 2012 to June 2013. Based on these events, we established a number of amplitude and phase scintillation distributions, which include scintillation-index-magnitude distributions, event-duration distributions, and event-occurrence frequency distributions. These results show very different characteristics of scintillation observed at low latitudes and high latitudes, indicating that there must be different mechanisms contributing to the formation and evolution of ionosphere plasma irregularities in the two regions. These characteristics are useful for scintillation-event prediction and modeling in the future.
Data Collection System and Event Thresholds
FIGURE 1 illustrates the general architecture of the event-driven GNSS data collection system. The system hardware consists of a multi-band GNSS antenna, a commercial ionospheric scintillation monitor (ISM) receiver, an array of reconfigurable software-defined radio (SDR) radio-frequency (RF) front-end devices capable of sampling intermediate-frequency (IF) signals, one or multiple data collection servers, a data storage array, timing signal distribution hardware to ensure both time and frequency consistency across all RF front ends and receivers, and network/communication devices that allow remote access of the receivers and servers to monitor the status of the hardware, to query recorded data, and reset and reconfigure the data collection system.
FIGURE 1. General architecture of the event-driven GNSS data collection system deployed at several high-latitude and equatorial sites since 2009.
Custom-designed space weather event monitoring and trigger software resides on the data collection and control server. The ISM receiver operates continuously to produce and record routine measurements such as I and Q channel accumulator outputs, pseudorange, carrier phase, Doppler frequency, C/N0, and scintillation indices, while the SDR RF front ends only temporarily store the latest one-minute worth of IF samples in each device’s circular buffer. Scintillation event thresholds are pre-determined based on analysis of baseline data collected at the same local site using the same hardware. The real-time event trigger software compares ISM receiver measurements with the pre-set event threshold. If the measurements exceed the thresholds, the contents of the circular buffers will be written to the data storage array until after the event subsides. These raw IF samples are then further post-processed using a wide range of receiver processing algorithms for analysis of scintillation features and robust receiver algorithm development.
The high-latitude GNSS receiver array at Gakona, was initially established in 2009 and has been continuously evolving into a four-antenna array capable of collecting GPS L1, L2C, and L5 and GLONASS L1 and L2 signal data until its recent relocation to and upgrade at Poker Flat Research Range, north of Fairbanks. Several publications have discussed the system setup, receiver signal processing of data collected by the system, and characterization of high-latitude scintillations based on analysis of the array outputs (see Further Reading). In this article, only the data collected using the commercial ISM receiver are discussed because this is the longest operating receiver at this site. The receiver outputs L1C/A signal intensity and carrier-phase measurements at a rate of 50 Hz and semi-codeless tracking results of L2P(Y) at 1 Hz.
Since 2011, several GNSS data collection systems have been deployed at low-latitude locations, including Hong Kong, Singapore, Peru, Ascension Island, and Puerto Rico. In this article, we use results from the ISM receiver at Jicamarca, Peru, close to the geomagnetic equator. FIGURE 2 shows the data-collection-system-setup block diagram at Jicamarca. The ISM receiver used in this location generates 100-Hz carrier-phase measurements and I/Q channel correlator outputs; the latter are further processed to generate 50-Hz signal-intensity measurements for GPS L1C/A, L2C, and L5 signals and GLONASS, Galileo, and BeiDou open signals. Seven SDR front ends driven by the same oven-controlled crystal oscillator (OCXO) signal from the ISM receiver sample GPS, GLONASS, Galileo, and BeiDou open signals. Preliminary results obtained from these and other low-latitude SDR data have been presented in several papers in the archived literature (see Further Reading).
FIGURE 2. Current multi-GNSS data collection system configuration at Jicamarca Radio Observatory in Peru. (GLO = GLONASS, BDS = BeiDou System, VPN = virtual private network, ISMET = ionospheric scintillation monitoring event triggering, RAID = redundant array of independent disks)
The raw carrier-phase and signal-intensity measurements obtained from the two ISM receivers at Gakona and Jicamarca were detrended, from which the two scintillation indices S4 and σφ were computed using Equations (1) and (2). In the two equations, I and φ stand for detrended signal intensity and carrier phase, respectively, and <·> represents the expected value that is essentially the average value over the interval of interest. In this study, the interval of interest was set to 10 seconds to most effectively highlight scintillation features based on evaluations of several different time intervals between 10 and 60 seconds.
(1)
(2)
As we mentioned earlier, the characterization of scintillation was carried out on the basis of scintillation events extracted from the raw data. After the evaluation of non-scintillation events and baseline indicators, a set of criteria has been established to extract interesting events through a semi-automated process from a large amount of data while keeping the number of selected events caused by non-scintillation factors (such as multipath and interference) low. A brief summary and explanations of the criteria are listed as follows:
The elevation angle mask is 30° to reduce multipath effects.
The thresholds for S4 and σφ are 0.15 and 15° respectively for data collected at Gakona.
For Jicamarca data, the thresholds are 0.2 and 15° respectively.
To exclude interference cases, the index value has to remain above the threshold value for a minimum of 30 seconds to qualify as a scintillation event.
An event detected within 5 minutes of the end of another event is combined as one event with the previous one.
Scintillations experienced by multiple satellite signals simultaneously are treated separately, and events experienced simultaneously for all visible satellites are further analyzed to ensure that they are not caused by interferences.
Carrier cycle slip/loss-of-lock detection and repair procedures are implemented to determine whether these cases are caused by scintillation or other factors.
It is important to note that the above criteria and procedures contain some degrees of arbitration, especially the last two, as they were applied based on visual inspections. These artificially imposed rules nevertheless are necessary for statistical analysis and comparison of scintillation observations.
Results and Discussion
In this section, we discuss the data sets we have collected and analyzed.
Available Dataset from Alaska and Peru. The ISM receiver at Gakona, started recording effective GPS data in August 2010. Environmental issues and human factors lead to a few intermittent data gaps during the more than three and a half years of data recording.
TABLE 1 lists monthly normal operation days and the percentage of time when data were collected. In all, the results presented in this article are based on approximately 3,000 scintillation events extracted from 657 days’ worth of data that was collected in a time span of 32 months.
Similarly, the number and percentage of days of effective data from Jicamarca, are summarized in Table 2. The dataset from this location runs from November 2012 until June 2013. Roughly 2,000 scintillation events have been extracted to enable statistical comparison of characteristics of scintillation observed in high- and low-latitude regions.
Scintillation Indicator Distributions. The magnitudes of the two scintillation indices, S4 and σφ , are often used to indicate the intensity of ionospheric scintillation, as their values directly reflect the disturbance rate of received power and carrier-phase measurements. Although there have been discussions regarding the suitability of σφ as a phase scintillation indicator, it is, nevertheless, a measure of the magnitude of carrier variations in a certain spectral range that are related to scintillation activities. In the absence of a commonly accepted new indicator for phase scintillation, we will use σφ in this study simply as a means to measure the phase fluctuations. FIGURE 3 compares the intensity distributions of amplitude and phase scintillation observed at the Alaska (square markers) and Peru (triangle markers) sites. MaxS4/σφ in the figures is the peak S4 or σφ value during an amplitude or phase scintillation event, which is a more practical indicator of scintillation impact on GNSS receivers.
FIGURE 3. Maximum S4 and σφ distributions of (a) amplitude and (b) phase scintillation observed at Gakona, Alaska, and Jicamarca, Peru.
Figure 3a shows that amplitude scintillation events observed at Jicamarca are generally more intense than those observed at Gakona. This is consistent with most previous studies, which concluded that scintillation is the most intense in the equatorial region. Figure 3b, on the other hand, shows that the intensity of phase scintillation at Jicamarca is slightly lower than that at Gakona. Nevertheless, this result does not necessarily reflect scintillation intensity observed in other parts of the equatorial region, as Jicamarca is not located close to the equatorial anomaly crest where scintillation activity is the strongest.
The duration of a scintillation event is another indicator of scintillation’s negative impact on the acquisition and tracking processes in receivers. FIGURE 4 plots the amplitude and phase event duration probability distributions, with the mean event durations at each site shown in the plots. The results show that at Gakona (square markers), phase scintillation lasts much longer than amplitude scintillation. At Jicamarca (triangle markers), amplitude scintillation events last slightly longer than the phase ones on average, and both types have much longer durations than those at high latitudes.
FIGURE 4. Duration distributions of (a) amplitude and (b) phase scintillation events observed at Gakona, Alaska, and Jicamarca, Peru.
Ionospheric scintillation of combined high intensity and long duration is usually considered a big threat to signal processing in GNSS receivers. Unfortunately, these two aspects are often correlated, especially at low latitudes. Moderate correlation coefficient values have been observed between scintillation durations and the magnitudes of scintillation indicators at Jicamarca (FIGURE 5b). The correlations, however, are much smaller at Gakona (FIGURE 5a), especially for amplitude scintillation events. These results further confirm that scintillation is a more severe issue in the equatorial region.
FIGURE 5. Scintillation duration vs. intensity at (a) Gakona, Alaska, and (b) Jicamarca, Peru.
Scintillation Occurrence Frequency and Relating Factors. We define the scintillation occurrence frequency as the number of scintillation events recorded during a certain time interval, which can be an hour, a day, a month, a season, and so on. The occurrence frequency is an important indicator in scintillation monitoring and forecasting, as it helps to identify the periods when scintillation events are most likely to occur.
FIGURE 6 illustrates scintillation hourly occurrence probabilities at the two sites with respect to Coordinated Universal Time (UTC) (upper) and hours post sunset (lower). Also consistent with numerous previous research findings, scintillation at high latitudes was more frequent during nighttime than at other times. Scintillation observed at Jicamarca occurred more frequently at night as well, but was greatly concentrated between one and two hours post sunset and midnight. Statistics show that 98% of Jicamarca’s scintillation events were observed from one to six hours after local sunset.
FIGURE 6. Scintillation occurrence frequency with respect to UTC hours and hours after sunset at (a) Gakona, Alaska, and (b) Jicamarca, Peru.
As demonstrated in Figure 6, scintillation occurrence frequency is largely influenced by solar inputs, which are the main driving force in atmospheric ionization and ionospheric irregularity formation. Scintillation occurrence can also be affected by geomagnetic activities. FIGURE 7 shows how scintillation occurrence frequency was affected by solar activity and seasons. The four seasons are defined as: spring (SP) – March to May; summer (SU) — June to August; fall (FA) — September to November; and winter (WI) – December to February. The intensity of solar activity is indicated by the smoothed average sunspot numbers, which are marked as black dots in the plot.
FIGURE 7. Seasonal scintillation occurrence frequency and smoothed sunspot number.
Several phenomena can be observed in Figure 7. At Gakona, scintillation occurrence frequency is clearly influenced by solar activity. The occurrence frequency is also modulated by season, with equinoxes generally more active than adjacent solstices. In contrast to the half-a-year cycle at high latitudes, scintillation occurrence frequency at Jicamarca more closely follows a one-year cycle as described in previous research, and decreases largely in the summer.
Our analysis also shows that the level of geomagnetic field activity also directly impacts scintillation occurrence frequency. FIGURE 8 shows the correlations between scintillation daily occurrence frequencies and Ap index values at the two sites. Ap is a widely used index that linearly reflects the daily average level of global geomagnetic field activity. Ap can be converted to the conventional Kp index using a quasi-logarithmic conversion table. The result in Figure 8a was obtained using data collected during seven months at Gakona: March and November 2011; March, July, October, and November 2012; and March 2013. During these months, scintillation activity was generally high. Figure 8b was generated using all the data listed in Table 2. Clearly shown in the plots, scintillation occurrence frequency at high latitudes is strongly correlated with geomagnetic field activities, while at Jicamarca such correlations do not exist. This result also confirms many previous research findings.
FIGURE 8. Daily scintillation occurrence frequency with respect to Ap index value at (a) Gakona, Alaska, and (b) Jicamarca, Peru.
Summary and Conclusions
This article presented comparative work on ionospheric scintillation characterization using data collected at Gakona, Alaska, and Jicamarca, Peru, during the current solar maximum to investigate the different natures of scintillation at high latitude and in equatorial regions. Scintillation intensity, duration, and occurrence frequency distributions were analyzed to demonstrate the differences at the two locations.
Scintillation in the equatorial region is typically more severe with deeper and faster signal power fadings and longer durations. Also, low-latitude scintillation with stronger intensity usually lasts longer, which further contributes to its negative impact on receivers. At high latitudes, phase fluctuations overwhelmed amplitude scintillation by the number of occurrences and their duration.
Scintillation is more frequent during nighttime, and almost all low-latitude scintillation events occur within six hours after local sunset. The overall occurrence frequency of scintillation not only increases with high solar activity, but also follows certain seasonal patterns. In general, scintillation is more active around the equinoxes. Additionally, high-latitude scintillation is also closely correlated to geomagnetic field activity, while the relationship is not obvious in the equatorial region.
Lastly, we would like to point out that the results presented here are preliminary and may be restricted to local effects, especially at low latitudes. As more data become available from Jicamarca and other equatorial sites where SDR data collection systems ensure quality inputs during strong scintillation events, a more comprehensive analysis and comparison can be made to facilitate global scintillation monitoring, mapping, and modeling.
Acknowledgments
The data collection and analysis project discussed in this article was supported by the U.S. Air Force Office of Scientific Research and Air Force Research Laboratory grants. The authors appreciate the support of High Frequency Active Auroral Research Program (HAARP) staff and the University of Alaska Fairbanks Geophysical Institute for organizing and sponsoring the HAARP campaign and HAARP staff support of the GNSS receiver data collection system setup. The authors would also like to acknowledge Jicamarca Radio Observatory for hosting the GNSS equipment. The Jicamarca Radio Observatory is a facility of the Instituto Geofisico del Peru, operated with support from the U.S. National Science Foundation through Cornell University. This article is based, in part, on the paper “Comparative Studies of High-latitude and Equatorial Ionospheric Scintillation Characteristics of GPS Signals” presented at PLANS 2014, the Institute of Electrical and Electronics Engineers / Institute of Navigation Position, Location and Navigation Symposium held in Monterey, California, May 5–8, 2014.
Manufacturers
The commercial ISM receivers used at Gakona and Jicamarca were a GPS Silicon Valley — now NovAtel Inc. — GSV4004B and a Septentrio N.V. PolaRxS Pro, respectively.
YU JIAO is a Ph.D. candidate at the Colorado State University (CSU), Fort Collins, Colorado. She received her master’s degree in computational science and engineering from Miami University, Oxford, Ohio, in 2013 and her bachelor’s degree in electronic and information engineering from Beihang University (previously known as the Beijing University of Aeronautics and Astronautics), Beijing, China, in 2011. Her research interests are in GNSS signal processing and ionosphere effects on GNSS in both high-latitude and equatorial regions.
YU (JADE) MORTON is an electrical engineering professor at CSU. She received a Ph.D. in electrical engineering from Pennsylvania State University (Penn State), State College, Pennsylvania, and was a post-doctoral research fellow in the Space Physics Research Laboratory of the University of Michigan, Ann Arbor, Michigan. Prior to joining CSU, she was a professor in the Department of Electrical and Computer Engineering at Miami University. Her research interests are advanced GNSS receiver algorithms for accurate and reliable operations in challenging environments, studies of the atmosphere using radar and satellite signals, and development of new applications using satellite navigation technologies.
STEVE TAYLOR is a graduate student in the Department of Electrical and Computer Engineering at Miami University. He received his B.S. in computer science from Miami University in 2011. Taylor developed software systems for ionosphere space weather monitoring and has been involved in deployment of Dr. Morton’s research team’s GNSS data collection system in Alaska, Peru, Hong Kong, Ascension Island, and Puerto Rico.
WOUTER PELGRUM is an assistant professor of electrical engineering at Ohio University, where he conducts research in and teaches about topics in electronic navigation, such as GNSS, Distance Measuring Equipment or DME, and time and frequency transfer. Before joining Ohio University in 2009, he worked in private industry, where he contributed to the development of an integrated GPS-eLoran receiver and antenna. From 2006 until 2008 he operated his own company, specializing in navigation-related research and consulting.
FURTHER READING
• Authors’ Conference Paper
“Comparative Studies of High-latitude and Equatorial Ionospheric Scintillation Characteristics of GPS Signals” by Y. Jiao, Y. Morton, and S. Taylor in Proceedings of PLANS 2014, the Institute of Electrical and Electronics Engineers / Institute of Navigation Position, Location and Navigation Symposium, Monterey, California, May 5–8, 2014, pp. 37–42, doi: 10.1109/PLANS.2014.6851355.
• Introduction to Ionospheric Scintillation and GNSS
“Limitations in GPS Receiver Tracking Performance Under Ionospheric Scintillation Conditions” by S. Skone, K. Knudsen, and M. de Jong in Physics and Chemistry of the Earth, Part A: Solid Earth and Geodesy, Vol. 26, No. 6-8, 2001, pp. 613–621, doi: 10.1016/S1464-1895(01)00110-7.
“Radio Wave Scintillations in the Ionosphere” — a review paper by C.K. Yeh and C.-H. Liu in Proceedings of the IEEE, Vol. 70, No. 4, 1982, pp. 324–360, doi: 10.1109/PROC.1982.12313.
High-Latitude Scintillations
“Characterization of High Latitude Ionospheric Scintillation of GPS Signals” by Y. Jiao, Y. Morton, S. Taylor, and W. Pelgrum in Radio Science, Vol. 48, 2013, pp. 698–708, doi: 10.1002/2013RS005259.
Equatorial Scintillations
“Statistics of GPS Scintillations over South America at Three Levels of Solar Activity” by A.O. Akala, P.H. Doherty, C.E. Valladares, C.S. Carrano, and R. Sheehan in Radio Science, Vol. 46, No. 5, October 2011, doi: 10.1029/2011RS004678.
“Measuring Ionospheric Scintillation in the Equatorial Region over Africa, Including Measurements from SBAS Geostationary Satellite Signals” by A.J. Van Dierendonck and B. Arbesser-Rastburg in Proceedings of ION GNSS 2004, the 17th International Technical Meeting of the Satellite Division of The Institute of Navigation, Long Beach, California, September 21–24, 2004, pp. 316–324.
“An Improved Ionosphere Scintillation Event Detection and Automatic Trigger for GNSS Data Collection Systems” by S. Taylor, Y. Morton, Y. Jiao, J. Triplett, and W. Pelgrum in Proceedings of ION ITM 2012, The Institute of Navigation 2012 International Technical Meeting, Newport Beach, California, January 30 – February 1, 2012, pp. 1563–1569.
Software Defined Radio Processing of GPS Scintillation Data
“Triple Frequency GPS Signal Tracking During Strong Ionospheric Scintillations over Ascension Island” by M. Carroll, Y.J. Morton, and E. Vinande in Proceedings of PLANS 2014, the Institute of Electrical and Electronics Engineers / Institute of Navigation Position, Location and Navigation Symposium, Monterey, California, May 5–8, 2014, pp. 43–49, doi: 10.1109/PLANS.2014.6851356.
Forecasting Scintillations
“A Forecasting Ionospheric Real-time Scintillation Tool (FIRST)” by R.J. Redmon, D. Anderson, R. Caton, and T. Bullett in Space Weather, Vol. 8, No. 12, December 2010, doi: 10.1029/2010SW000582.
“Specification and Forecasting of Scintillations in Communication/Navigation Links: Current Status and Future Plans” by S. Basu, K.M. Groves, Su. Basu, and P.J. Sultan in Journal of Atmospheric and Solar-Terrestrial Physics, Vol. 64, 2002, pp. 1745–1754, doi: 10.1016/S1364-6826(02)00124-4.
Alternative Scintillation Indices
“Improved Amplitude- and Phase-scintillation Indices Derived from Wavelet Detrended High-latitude GPS Data” by S.C. Mushini, P.T. Jayachandran, R.B. Langley, J.W. MacDougall, and D. Pokhotelov in GPS Solutions, Vol. 16, No. 3, July 2012, pp. 363–373, doi: 10.1007/s10291-011-0238-4
“Perils of the GPS Phase Scintillation Index (sf)” by T.L. Beach in Radio Science, Vol. 41, RS5S31, 2006, doi: 10.1029/2005RS003356.
“Problems in Data Treatment for Ionospheric Scintillation Measurements” by B. Forte and S.M. Radicella in Radio Science, Vol. 37, No. 6, 1096, 2002, pp. 8-1–8.5, doi: 10.1029/2001RS002508.
