An approach for ground vehicles using carrier-phase and inertial measurement data
The combination of easily accessible low-cost GNSS spoofers and the emergence of increasingly automated GNSS-reliant ground vehicles prompts a need for fast and reliable GNSS spoofing detection. To underscore this point, Regulus Cyber, an Israeli cybersecurity company, recently spoofed a Tesla Model 3 on autopilot mode, causing the vehicle to suddenly slow and unexpectedly veer off the main road.
Among GNSS signal authentication techniques, signal-quality monitoring (SQM) and multi-antenna could be considered for implementation on ground vehicles. However, SQM tends to perform poorly on dynamic platforms in urban areas where strong multipath and in-band noise are common, and multi-antenna spoofing detection techniques, while effective, are disfavored by automotive manufacturers seeking to reduce vehicle cost and aerodynamic drag. Thus, there is a need for a single-antenna GNSS spoofing detection technique that performs well on ground vehicles, despite the adverse signal-propagation conditions in an urban environment.
In a concurrent trend, increasingly automated ground vehicles demand ever-stricter lateral positioning to ensure safety of operation. An influential study calls for lateral positioning better than 20 centimeters on freeways and better than 10 centimeters on local streets (both at a 95% probability level). Such stringent requirements can be met by referencing lidar and camera measurements to a local high-definition map, but poor weather (heavy rain, dense fog or snowy whiteout) can render this technique unavailable.
On the other hand, progress in precise (decimeter-level) GNSS-based ground vehicle positioning, which is impervious to poor weather, has demonstrated surprisingly high (above 97%) solution availability in urban areas. This technique is based on carrier-phase differential GNSS (CDGNSS) positioning, which exploits GNSS carrier-phase measurements having millimeter-level precision but integer-wavelength ambiguities.
Key to our promising results is the tight coupling of CDGNSS and inertial measurement unit (IMU) data, without which high-accuracy CDGNSS solution availability is significantly reduced due to pervasive signal blockage and multipath in urban areas. Tight coupling brings millimeter-precise GNSS carrier-phase measurements into correspondence with high-sensitivity and high-frequency inertial sensing. Our particular estimation architecture incorporates inertial sensing via model replacement, in which the estimator’s propagation step relies on bias-compensated acceleration and angular rate measurements from the IMU instead of a vehicle dynamics model.
As a consequence, at each measurement update, an a priori antenna position is available whose delta from the previous measurement update accounts for all vehicle motion sensed by the IMU, including small-amplitude high-frequency motion caused by road irregularities. Remarkably, when tracking authentic GNSS signals in a clean (open-sky) environment, the GNSS carrier-phase predicted by the a priori antenna position and the actual measured carrier phase agree to within millimeters.
The research described in this article pursues a novel GNSS spoofing-detection technique based on a simple but consequential observation: it is practically impossible for a spoofer to create a false ensemble of GNSS signals whose carrier-phase variations, when received through the antenna of a target ground vehicle, track the phase values predicted by inertial sensing. In other words, antenna motion caused by factors such as road irregularities or rapid braking or steering is sensed with high fidelity by an onboard IMU but is unpredictable at the sub-centimeter-level by a would-be spoofer.
Therefore, the differences between IMU-predicted and measured carrier-phase values offer the basis for an exquisitely sensitive GNSS spoofing-detection statistic. What is more, such carrier-phase fixed-ambiguity residual cost is generated as a byproduct of tightly coupled inertial-CDGNSS vehicle position estimation.
Two difficulties complicate the use of fixed-ambiguity residual cost for spoofing detection. First is the integer-ambiguous nature of the carrier-phase measurement, which causes the post-integer-fix residual cost to equal not the difference between the measured and predicted carrier phases (as would be the case for a typical residual), but rather modulo an integer number of carrier wavelengths. Such integer folding complicates development of a probability distribution for a detection test statistic based on carrier-phase fixed-ambiguity residual cost.
Second, the severe signal multipath conditions in urban areas create thick tails in any detection statistic based on carrier-phase measurements. Setting a detection threshold high enough to avoid false spoofing alarms caused by mere multipath could render the detection test insensitive to dangerous forms of spoofing. Reducing false alarms by accurately modeling the effect of a particular urban multipath environment on the detection statistic would be a Sisyphean undertaking, requiring exceptionally accurate up-to-date 3D models of the urban landscape, including materials properties.
Our work takes an empirical approach to these difficulties. It does not attempt to develop a theoretical model to delineate the effects of integer folding or multipath on its proposed carrier-phase fixed-ambiguity residual cost-based detection statistic. Rather, it develops null-hypothesis empirical distributions for the statistic in both shallow and deep urban areas, and uses these distributions to demonstrate that high-sensitivity spoofing detection is possible despite integer folding and urban multipath.
Anticipating New, Different Application and User Needs
Users in emerging applications may have different requirements from traditional high-precision users. New users increasingly look to the technology not solely for position, but to navigate them through the environment, often autonomously or semi-autonomously. Tracking all of the new multi-GNSS signals, and then using the large number of inputs in the positioning engine, drives the amount of processing power and memory required onboard the receiver. These in turn drive the cost, size and power consumption of the receiver in exactly the opposite direction from the expectations of customers.
By Jason Hamilton
In considering the future of high-precision satellite navigation, we need to consider what users of the technology are trying to accomplish, and which growing and emerging applications will drive adoption of GNSS technology in the future. These applications will drive growth in our industry if we can correctly anticipate their future needs.
Traditional applications of high-precision GNSS are well understood, but what these customers have demanded from GNSS can be at odds with what users in emerging applications require. Survey and mapping users were early adopters of high-precision GNSS and remain large user segments. Surveying with GNSS requires the very best accuracy that GNSS can achieve. Every centimetre of accuracy matters. Power and size are important product attributes to survey manufacturers. Mapping customers increasingly are asking for not just position, but orientation of a camera or other sensors.
Once accuracy challenges were well in hand, the topic of availability came into play. It was no longer good enough to have an accurate position in open-sky situations. Applications demanded continuous positions that were accurate in more and more corner cases and challenging environments.
In addition to using GNSS to measure location in an environment, new applications are increasingly looking to the technology to navigate them through the environment — often autonomously, or semi-autonomously. For these users, whether operating on a farm, in a mine, on the ground, or in the air, position accuracy is only part of the requirement. Solution accuracy of course matters, but other receiver attributes such as real-time quality control and solution integrity monitoring, are equally or more important.
Multi-constellation, multi-frequency GNSS provides tremendous opportunity and also presents significant challenges for receiver manufacturers. Constellation and frequency support has previously been a differentiator among high-precision GNSS providers, and among product generations. The relative stability of the satellite constellation definition means that the signals broadcast from space will be relatively predictable for some time into the future, and as such, GNSS products are increasingly supporting “all in view,” the ability to track everything that is broadcast.
The benefits of more satellites, more frequencies (and resulting frequency combinations) and modern signal structures have been well publicized. As new and modernized GNSS constellations come on line, they will deliver more robust positioning in increasingly challenging environments such as urban centers, open-pit mines and under tree cover. We will be able to account for atmospheric effects more accurately, which will help during times of high ionospheric activity and extend the length of RTK baselines. Users have a great deal to look forward to from their next-generation receivers.
All of these improvements necessitate pretty dramatic changes in receiver design. Tracking four global constellations and numerous regional SBAS systems increases the complexity of tracking and positioning firmware and algorithms. Tracking multiple frequencies and signal types on each of these constellations drives the receiver channel count up substantially. The days of the 12-channel receiver are gone. Channels, typically implemented within the manufacturers’ custom chips, drive application-specific integrated circuit (ASIC) complexity, which drives cost, power consumption and physical size. Some of this can be mitigated through the use of smaller process geometries, embedded processors and peripherals, and RF chip integration; however, there are down-stream effects to all of these signals as well.
Challenges
Once your receiver has enough ASIC channels to track all-in-view, you need to do something with all that data. The receiver’s tracking sub-system generates code (pseudorange), carrier-phase and Doppler measurements for every signal on each satellite. With four global and multiple regional constellations and up to four frequencies on each satellite, that amounts to a great deal of data. These measurements are what we turn into position, through a range of different positioning algorithms from code positioning to real-time kinematic (RTK) to precise point positioning (PPP). Tracking all of these signals, and then using the large number of inputs in the positioning engine, drives the amount of processing power and memory required onboard the receiver. These in turn drive the cost, size and power consumption of the receiver in exactly the opposite direction from the expectations of customers.
Bandwidth. Communications bandwidth is also a future challenge. Positioning methods, such as RTK, that transmit base-station observations for each GNSS signal to field rover receivers, will require much more bandwidth in the all-in-view future. PPP, which provides a state-space correction of the underlying GNSS error sources, is a promising alternative to RTK that scales better with more satellites than RTK and provides performance that is good enough for many applications.
Utilizing the multiple frequencies available from modern constellations also presents challenges to receiver designers. RF designers are faced with the opposing challenges of making GNSS receivers and antennas smaller, lighter and lower cost, while also supporting more GNSS broadcast frequencies and mitigating against increasing amounts of interference in the L-band RF spectrum from non-GNSS uses. Robust RF design makes the difference between a system that works most of the time, and a system that works reliably all of the time.
Expectations
If we now come back to the expectations of end users, the challenges are clear. Most customers actually don’t care about all-in-view tracking, how many satellites are tracked, or about what the receiver is up to behind the scenes. Users will judge their GNSS receiver on whether or not they are receiving a position that meets the requirements of their application. Are they meeting their targets for accuracy, availability, latency, data rate, and does the receiver fit from a size, power consumption, regulatory and cost perspective? After a certain level, more observations do not make the solution more accurate or more robust. Manufacturers need to carefully manage the tradeoffs in their systems on behalf of users to produce the best quality position possible, while still meeting the customer expectations on all the other receiver attributes.
Sensor Fusion. Demands of new applications drive GNSS providers to consider more than just position. Most vehicle control applications require orientation information as well as highly accurate position. Multiple-antenna GNSS heading systems are becoming smaller than ever. Inertial measurement device technology is also evolving quickly. Miniature micro-electro-mechanical systems (MEMS) inertial sensors can now deliver performance that only a few years ago was exclusive to large, heavy, bulky systems. The integration of GNSS and inertial technologies has been well adopted in highly demanding applications like aerial and ground mapping. As the size, weight and cost of the technology continues to shrink, sensor fusion in many forms will become the standard for all machine control and autonomous vehicle applications.
Safety. This is a key consideration for system designers working on remotely or optionally piloted and autonomous systems. Position and orientation accuracy is important, but so, too, is assuring that the solution is right and can be trusted. The accuracy of the solution needs to be characterized in real time so that control systems can react as necessary to protect users on and around the vehicle. Often in these applications, accuracy can be traded off against the robustness and reliability of the solution. This presents new ways of thinking for firmware and algorithm developers who have focused for so long on solution accuracy.
Support. Lastly, let’s not forget having reliable supply of high-quality product, and expert customer service to back it up. As high-precision GNSS attracts new users in a range of new industries, they are less often geodesists or geomatics engineers. The products absolutely need to be easy to use correctly, backed up by complete and accurate product documentation and supported by world-class application engineers.
Jason Hamilton is vice president of marketing at NovAtel Inc. Since joining the company, he has held a number of research, development and product management roles. Jason holds a Bachelor of Science degree in geomatics engineering from the University of Calgary and an MBA from Royal Roads University.
SVN49 in space (artist’s rendering). The signal anomaly from SVN 49 alerted researchers to new possibilities in analysis and monitoring.
Chip Transition-Edge Based Signal Tracking for Ultra-Precise GNSS Monitoring Applications
By Sanjeev Gunawardena, John Raquet and Frank van Graas
Tracking GNSS signals using their underlying spreading sequence chip transition edges reveals positive versus negative chip asymmetries that are characteristic to each satellite. This asymmetry is due to various types of natural signal deformation that is known to occur within the satellite’s signal generation and transmission hardware. This novel concept of monitoring chip asymmetry can extend the state of the art in the areas of GNSS signal-quality monitoring and authentication. A technique to directly monitor chip asymmetry within a specially designed ChipShape GNSS receiver architecture employs separate code discriminators that align themselves to the chip rising-edge and falling-edge zero crossings.
The detailed study of naturally-present deformations in GNSS signals is a relatively new activity that was sparked by the GPS SVN49 anomaly and the associated research activities that followed. This research area has numerous applications that include:
Informing the design of sudden signal deformation detection and alerting algorithms for safety-of-life differential GNSS applications (such as aviation).
GNSS signal “fingerprinting” and authentication.
The detailed study of long-term degradation effects of GNSS satellite signal generation and transmission hardware.
Analysis of the impact to the first item in this list of swapping a satellite’s signal generation modules by its control segment.
Multipath detection, characterization, and mitigation are also closely tied to all research relating to GNSS signal deformation monitoring (SDM).
High-fidelity SDM can be performed using two methods:
observation of actual GNSS signals above the thermal noise floor using a high-gain dish antenna;
the combination of long coherent integration and multi-correlator processing.
Our previous research has revealed that these two methods are highly complementary for gaining full insight into the effects and causes of observed natural signal deformations.
Among the handful of multi-correlator processing techniques that can be applied for SDM, ChipShape processing allows the correlation function resolution to be finely adjustable while providing good numerical processing efficiency. This processing technique also allows chip-transition eye diagrams to be constructed in order to provide additional insight such as positive and negative chip width asymmetries.
One goal of our SDM research involves developing capabilities to observe GNSS signals with the highest levels of fidelity practically achievable in order to further the application areas described above. Key to this is developing techniques to track GNSS signals using a reference point that is both consistent and invariant (to the greatest extent possible) to nominal signal deformations and environmental effects such as multipath. Traditional multipath mitigating techniques such as narrow correlator and double-delta correlator are sub-optimal in this regard. This is because a significant portion of the signal around the chip transition point (that is, 10 percent and 20 percent for 0.1 chip correlator spacing, respectively) must be integrated to realize these discriminators and maintain robust tracking in moderate dynamics conditions. This integration tends to low-pass filter the desired observables.
Chip Transition Edge-Based Code Tracking
Figure 1 illustrates normalized C/A code chip rising edges for the GPS constellation of June 2014. These chip shapes were processed using a front-end with 24 MHz bandwidth. For visual comparison purposes, this and other related plots were obtained using 600 seconds of coherent integration.
Figure 1. Normalized ChipShape rising edges for the GPS SPS constellation of June 2014; each color represents a different GPS satellite.
The code tracking loop used to obtain this result employed an empirical normalized coherent rising-edge discriminator given by:
(1)
Where τ is relative code phase in chips, d is Early-Late correlator spacing,R’XYZ(i) is the differential correlation output for integer bin i obtained using ChipShape processing with masking sequence XYZ. bin(x) is a function that selects the closest ChipShape vector index that corresponds to relative code phase x. Each ChipShape processing bank is configured to span one chip early and one chip late with a resolution of N bins per chip, thus producing a ChipShape vector of 3N bins. α is a scale factor obtained through trial and error to yield robust tracking performance as observed by the code-minus-phase measurement. For the result shown in Figure 1, N=240 and d ≈ 0.017 chips.
The figure clearly shows that the rising-edge zero crossings vary by SV. This variation is due to nominal signal deformation present in each GPS-SPS signal.
Figure 2 illustrates the rising-edge zero crossings aligned to zero relative code phase. This alignment was performed by interpolating each R’NPN vector, precisely estimating code phase at the zero-crossing point, and shifting the curve appropriately.
Figure 2. Normalized ChipShape rising edges for the GPS SPS constellation of June 2014: Zero crossing compensated.
Figure 3 shows zero crossings for the falling edges after all rising edges were aligned to zero. The figure clearly illustrates subtle asymmetries between positive and negative chips which span a range of approximately ±1.5 meters. These asymmetries are not directly observable using typical GNSS receiver processing. However, they can lead to pseudorange biases through the resulting distortion that occurs to the traditional correlation function.
Figure 3. Normalized ChipShape falling edges for the GPS SPS Constellation of June 2014 when rising edges are aligned to zero.
In general, a family of code discriminators that precisely track chip rising-edge zero crossings can be defined by:
(2)
Where R’NPX is a linear combination of orthogonal ChipShape components that preserve the rising-edge transition, e.g.: R’NPX =R’NPN +R’NPP. R’FFX is a linear combination of orthogonal ChipShape components that preserve the non-transitioning (that is, flat) sections of chips, for example: R’FFX =R’PPP + R’PPN −R’NNP − R’NNN. a and b define an integration interval within the range −1 to +2 chips with respect to the chip transition edge. β is a bias compensation term. represents the real or imaginary component function for the coherent discriminator (depending on the modulation phase of the signal being tracked), or the magnitude function for a non-coherent discriminator implementation.
Similarly, a family of code discriminators that precisely track chip falling-edge zero crossings that occur one chip after the rising edges tracked by the discriminator of Equation 2 can be defined by:
(3)
Then, a two-step technique to precisely monitor chip asymmetry can be described as follows:
Setup two identical ChipShape processing channels to track a given PRN. Progressively tighten the code tracking loops to track the rising-edge zero crossings of the underlying signal using the discriminator of Equation 2.
After steady-state zero-crossing rising-edge tracking is achieved, switch the second channel’s code discriminator to that of Equation 3. This will cause the second channel to track the zero crossings of the falling edges that occur one chip later in the underlying signal’s spreading sequence. The discriminator’s linear range must be wide enough to pull-in the chip asymmetry shown in Figure 3.
When the second channel re-converges as a result of Step 2, the relative pseudorange displacement that occurs is equal to the chip asymmetry in meters. Hence, chip asymmetry can be monitored for the entire visible pass of a satellite. It is expected that positive and negative chip transitions are equally affected by channel distortions (that is, code and carrier multipath, ionosphere, troposphere, and the receiver antenna and front-end transfer function). Hence, the rising-edge-code-minus-falling-edge-code measure of chip asymmetry is expected to be invariant to most if not all channel distortions.
Estimating Compensation Parameters
As shown in Equations 2 and 3, due to natural signal deformation of many types, the rising and falling-edge zero-crossing discriminators are expected to be SV number, PRN code and elevation angle dependent. Hence, α and β must be estimated for a given correlator spacing d separately for all SV signals of the constellation. These values will also be specific to a given antenna and receiver front-end.
Figure 4 illustrates the procedure used to estimate the scale factor and bias terms starting with the empirical rising-edge tracking process described above.
Figure 4. Procedure for estimating scale factors and biases for rising-edge tracking early-late and double-delta code discriminators.
The following figures illustrate the edge tracking discriminator calibration process using R’NPN for a single SV.
Figure 5 illustrates the early-plus-late functions computed for various correlator spacings. As described previously, these functions typically do not cross through zero codephase due to natural signal deformation.
Figure 5. Uncorrected rising-edge early-late discriminator functions for various correlator spacings.
Figure 6 illustrates the rising-edge discriminator functions after bias compensation.
Figure 6. Rising-edge early-late discriminator functions for various correlator spacings after bias compensation.
Figure 7. Calibrated rising-edge early-late discriminator functions for various correlator spacings.
Figure 8 illustrates the multipath error envelopes for the rising edge-based coherent code discriminators. The performance of these discriminators is similar to the traditional Early-Late discriminators for the same correlator spacings. This result is consistent with the theoretical bounds for code multipath.
Figure 8. Multipath error envelopes for various rising edge-based coherent early-late code discriminator functions.
As shown in Figure 4, the edge-tracking discriminators described in Equations 2 and 3 that are based on Early-Late bin spacings can be combined to obtain edge-tracking double-delta discriminators. Double-delta discriminators provide significantly improved multipath performance.
In general, the edge-tracking double-delta discriminator for inner correlator spacing d is formed by the linear combination of two early-late edge-tracking discriminators, as follows:
(4)
Scale factor γ is estimated such that overall multipath error is minimized according to a given design criteria.
Figure 9 illustrates the double-delta rising-edge discriminator with inner spacing of 0.017 chips. This discriminator has a pull-in range of approximately ±0.01 C/A chips.
Figure 11 illustrates the multipath error envelope for the coherent rising-edge double-delta discriminator. Performance is consistent with a traditional second-derivative discriminator.
Figure 11. Multipath error envelope for coherent rising-edge double-delta code discriminator with inner spacing of ~0.017 C/A chips.
Figure 12 illustrates the performance of the various rising-edge tracking discriminators for a live-sky GPS-SPS signal (de-trended code-minus-carrier measurement). This figure clearly demonstrates robust code tracking and the multipath and noise mitigating benefit of ultra-narrow rising-edge discriminators.
Figure 12. Code tracking performance for live sky data of various rising edge-based coherent early-late code discriminator functions.
Conclusions
An empirical chip rising edge-based tracking technique was used to observe the underlying chip shapes of live sky GPS-SPS signals at high fidelity. These results reveal positive versus negative chip asymmetries that are characteristic to each satellite. The novel concept and technique of directly monitoring chip asymmetry has potential to extend the state of the art in the areas of GNSS signal quality monitoring and authentication.
Disclaimers. The views expressed in this paper are those of the authors and do not reflect the official policy or position of the United States Air Force, Department of Defense, or the United States Government.
Acknowledgments. This research was supported by the Air Force Research Laboratory Sensors Directorate.
The authors thank Ohio University Avionics Engineering Center for making available a cluster of high-performance computers to process the 20 TB dataset for this research, and Kadi Merbouh of Ohio University for maintaining and overseeing operation of this equipment.
The ChipShape processing is an extension of the signal compression technique first published by Larry Weill and licensed by NovAtel for use in its Vision Correlator technology.
This article is based on a paper presented at ION Pacific PNT 2015 in Honolulu.
SANJEEV GUNAWARDENA is a research assistant professor with the Autonomy & Navigation Technology (ANT) Center at the Air Force Institute of Technology (AFIT). He earned a Ph.D. in electrical engineering from Ohio University.
JOHN RAQUET is a professor of electrical engineering and the Director of the ANT Center at AFIT. He has been involved in navigation-related research for more than 25 years.
FRANK VAN GRAAS is the Fritz J. and Dolores H. Russ professor of electrical engineering and principal investigator with the Avionics Engineering Center at Ohio University. He received the ION Johannes Kepler, Thurlow and Burka awards, and is a Fellow and past president of the ION.
