Resilient PNT: How to Detect, Protect and Prevent Disruption of GPS/GNSS/PNT Signals and Sources

For the last two decades, reliance upon GPS/GNSS for positioning, navigation and timing (PNT) by civilian applications has been an overwhelming success, providing increasingly accurate and reliable PNT information everywhere for free. However, in recent years, malicious threats have emerged have exposed the Achilles heel of GPS/GNSS: Its weak power and unencrypted signal format leave it vulnerable to jamming and spoofing. The loss of PNT information can result in rare but catastrophic events, crippling critical infrastructure that is necessary for our survival.

Fortunately, there are many alternative technologies available to augment GPS and provide accurate and reliable PNT information even under extremely adverse conditions. These include:

• Low Earth Orbit (LEO) PNT satellite signals – operating close to the Earth, with signal strengths ~1000x stronger than GPS, to overcome jamming. And, encrypted to prevent spoofing.
• Miniaturized, affordable atomic clocks to provide precise time internally – which is the foundation of every positioning system –, without connection to, or reliance on, outside sources which could corrupt the precision.
• Fiber optic network time distribution via secure, high accuracy protocols.

In this white paper, we will discuss these alternative technologies and offer recommendations for procedures and processes to manage and avoid the risks of cyberattacks, applying the best practices known today. We will focus on Resilient PNT and describe the strategies, technologies and procedures to avoid these cybersecurity risks and ensure the continuity of any critical operation.

• GNSS vs. GPS – all GNSS systems (GLONASS, Galileo, Beidou, and GPS) are very similar in operation and have similar vulnerabilities and strengths, so using multi-constellations does not offer much diversity. They all share the same few frequency bands. In general, if one system is jammed, they all are. However, using multiple constellations will yield independent verification of each system. It takes a more sophisticated spoofer to fake all constellations, but it can be done.

The other advantage of using multi-constellations is that it will catch an anomaly on an individual system. For example, the 13-microsecond GPS glitch in January 2016, the error in the GLONASS system in April 2014, and the Galileo timing glitch in the summer of 2019, would all have been detected by systems using multi-GNSS receivers with appropriate cross-checking algorithms. On the other hand, these other systems are quite new and their failure modes are still unknown. Further, dependency on non-USA sources has national security issues too.

• LEO PNT – the broad applicability stated here is based not only on what is available today from STL transmitting from the 66 satellite Iridium constellation, but also includes the eventual performance expected when LEO PNT signals are available on thousands of satellites from other constellations such as OneWeb, Boeing, Space-X, etc. The accuracy limitations of STL today are due to two major factors: geometric dilution, because not enough satellites are visible simultaneously, and limited signal bandwidth. Neither of these two factors should be an issue for future systems.
• Dedicated Terrestrial Transmitters – includes technology represented by NextNav LLC, Locata, Phasor Lab and others. These systems require installing a transmitting infrastructure, so are best used in specific environments such as urban centers, warehouses or test ranges. Area of coverage depends on the allowed transmitter power.
• eLoran – does not exist operationally today in the USA, but it is applicable for a number of sectors if a largescale build-out were implemented. Prior to the existence of GPS, LOng RAnge Naviation (LORAN) was used for coastal and river marine navigation in the USA. A few dozen high power stations provided this coverage until the early 2000s. With today's Digital Signal Processing (DSP) technology, better accuracy and coverage is possible. With fewer than 100 stations, complete coverage of USA populated areas should be possible; even fewer would be needed for timing applications only.
• VOR/TAC/DME/ILS – VHF Omnidirectional Range, Tactical Aircraft Control, Distance Measuring Equipment, and Instrument Landing Systems are all ground-based RF signals generated near and around airports for flight guidance. One can imagine these old systems being adapted for other uses, but the signal structures were
  developed in the early to mid-20th century for primitive wireless electronics, so much better approaches can be imagined. However, changes to these signal formats is encumbered by the massive avionics installed base, to which compatibility must be maintained.

Signals of Opportunity are RF signals that were not designed for transmitting PNT information, but with little or no modification can be used to determine position or time.

• Cellular includes 4G/LTE and 5G networks. Positioning from cellular is done today, achieving better than 100m accuracy. 5G can provide higher accuracy than 4G/LTE,
  and further improvements can be considered so that position accuracy could approach GNSS, but as of today, it is less accurate. Not included here is how cellular provides
  the "Assisted GNSS" function – using the data network to transport navigation messages to GNSS receivers faster and more reliably than receiving them directly from the satellite.

• WiFi is neither accurate nor provides enough practical coverage to be considered for anything but a pedestrian use case – typically for indoor positioning.
• Broadcast TV assumes that improvements would be made to synchronize existing transmissions, but further build-out would most likely not happen in the future
  because the bandwidth is far too valuable for 5G. Any specific build-out of this technology approach for PNT is related to the high-power case of a Dedicated Terrestrial Transmitter signal described previously.
• RFID provides positional proximity indication and therefore is not useful in timing applications.

• Crowd-Sourced/Collaborative methods work on the idea that if other nodes on a network to which you are connected know their positions, and you have some indication of your relative proximity to them, you can infer your own position. Indications of proximity include RF signal strength on a wireless network, or roundtrip packet delay on a wired network. Analogously for the time sync situation, multiple nodes on a network can consensually sync together without necessarily being connected to a master source like GPS. The more participants on the network, the better the probability of resulting in a PNT solution, and the smaller the error ellipse will be. Of course, collaboration is not a PNT source itself, but when combined with diverse sources, it is a powerful estimation method.

Vision systems provide positional awareness much the same way as humans are made aware of their locations – through visual cues to recognized landmarks. However, they do not help with time determination.

• Camera imaging combined with map matching provides both position and attitude determination.
• Lidar uses laser range finding to create a 3D point map of a surrounding area.
• Radar operates similarly to lidar but with much less angular resolution. It is less impacted by smoke/fog/rain/snow than optical methods, and recent advances in semiconductors have made it very affordable.
• Sonar is not viable for high speed aircraft, but it provides proximity detection for use with other methods.

Sensing is like Vision but with these unique characteristics:

• Sensing the Earth's magnetic field is not a very accurate method for heading determination, but it can provide an approximate initial state for inertial navigation.
• Celestial navigation is an ancient form of navigation for mariners that has been automated with cameras and image recognition processing. However, stars are only visible on clear nights or at extremely high altitudes.
• Sky Polarization sensing takes advantage of the phenomenon where the sun's light is polarized differently at various angles of incidence with the atmosphere. By knowing the date and time and a rough estimate of your latitude and longitude, you can obtain a precise estimate of north orientation. Combined with celestial navigation, this can provide compass direction in both day and night conditions, though it does not operate well under all sky conditions with cloud cover.

Figure 2: Cellular Wireless Networks and PNT

current

1. Adjacent cell time and frequency sync
2. Assisted GNSS – nav msgs, time of day, and gross location received over cellular network (Mbps) instead of GNSS channel (50 bps) to shorten GNSS acquisition time
3. e911/eCall – cellural network provides ~100 M positioning required by law through combination GNSS and sector location/time of ariival

• Datacenters – The primary need for PNT here is to synchronize each processing element to a common time reference. This is necessary so that each element knows which transaction happened first, which happened second, etc. Examples of Datacenter operators are Google, AWS, Azure, and Verizon. In most cases, all that is needed is relative sync, not absolute sync, to Coordinated Universal Time (UTC). However, when processing elements are separated across wide areas (such as different cities or countries), GPS sync is the simplest and most accurate way to achieve it. Sync accuracy determines the efficiency of multi-processing. The more uncertain two processing elements are about their relative time references, the more "guard" or "wait" time they must use to have transactions settle before continuing or risk the probability of an error.

Figure 3: Datacenter Sync Application

A secondary need for PNT is for mobile data. As more transactions are executed from mobile users and, with the emergence of the Internet of Things (IoT), more objects that physically move are executing transactions. These data packets need to be accurately geo-timestamped for meaningful processing to occur. The quality of information extracted from "Big Data" mining is very dependent upon the veracity of the PNT tag.

With 5G comes more bandwidth, enabling more real-time applications to run in the cloud. With this trend, latency and time performance monitoring will become more critical.

• The Power Grid depends on precise time synchronization, as shown in Figure 3. Keeping the 60 Hz AC power in phase across the grid is a challenge. The main device supporting this process is the PMU – Phase Measurement Unit. It monitors the sine wave against a reference time, usually GPS, and provides feedback to the systems controlling phase alignment. One degree of phase error in this 60 Hz sine wave is 46 microseconds, so accuracies of tens of microseconds in measurement are desired, though ~100s microsecond errors can be tolerated. The other class of devices requiring sync are the Digital Fault Recorder (DFR) and Sequence of Event Recorders (SER). To diagnose faults and determine causal relationships, you must have a common time reference to know what happened first, what happened next, etc. Just like many other sectors, the power grid only needs relative time, not absolute UTC time, but when operating over wide distances, UTC or GPS time is the easiest way to access a common clock. Two trends are occurring in the power industry that will influence its evolving need for resilient PNT:
  1. Distributed power generation – as we move from central power generation from coal, oil, gas and nuclear stations to renewables like wind and solar, the sources of power are more distributed, more dynamic, and less controllable than before. This adds complexity to the phase controlling algorithms requiring more measurement points and stricter compliance to accuracy and reliability.
  2. Smart Grid – a data network is being superimposed on the grid so that every device on the grid has network connectivity. For these millions of devices to be managed and to interact with each other, they need PNT information to know where these devices are and for them all to have a common time reference.
     

After a dependent infrastructure is set up with multiple, diverse PNT sources, the best and simplest method for detection is to look for disagreement among the sources. With diverse sources, each one will have different points of vulnerabilities and different failure modes, so that failures or attacks will manifest in each source differently. Consider a timing system with three distinct sources: GNSS receiver, atomic clock, and PTP WR fiber network time. Jamming or spoofing of the GNSS signal will cause disagreement with the atomic clock and PTP-derived time; failure of the atomic clock will be detected by disagreement with GNSS and PTP time; hacking of the network with a Denial of Service (DoS) or Packet Delay attack will be detected by the GNSS and atomic clock. There is no single point failure.

Similarly, for a positioning example, consider a vehicle with a GNSS receiver, an IMU, and an STL receiver. Jamming of the GNSS receiver would normally force the tracking to be open loop, relying on the IMU only, which would drift over time. However, with updates
from the STL signals, which are 1000x stronger than GPS and more resistant to jamming, the error drift is indefinitely limited to tens of meters.

Spoofing of the GNSS signal is detected either by disagreement with STL, which is an encrypted signal that cannot be spoofed, or by the IMU, which does not sense movement in the false direction. Failure of any one of these three sources is noted by the agreement of the other two. Majority wins.

Each individual PNT source has its own detection and prevention measures. For example, there are many jamming and spoofing detection algorithms that can be applied to GNSS receivers to alert and often forewarn the infrastructure system of a possible disruption. False alarming is an issue, so using a probability-based scoring system with programmable thresholds can be helpful to keep both the false alarm and the missed detection rates low.

A simple GNSS spoofing detection technique is to use two separate receivers and antennas, spaced apart by a known distance. For example, placing one receiver/antenna combination at the fore of a ship, the other aft, will always yield two different position indications under normal operating conditions, based on their separation. If they ever indicate the same position, it is an indication of spoofing from an external source.

Detecting dropouts, discontinuities or other anomalous behavior from a particular sensor is another method of alerting. In our universe, it is impossible for an object to "jump" across the time-space continuum; it must move smoothly from one point in space-time to another and not exist at more than one point at any single instance. Therefore, an indication of discontinuity – within the bounds of measurement noise and quantization effects –
is a potential failure or false manipulation of the source. Discontinuity, measurement range checking and sensor behavior monitoring can be powerful detection methods.

• Vulnerability Testing – PNT systems should be analyzed and tested before they become part of critical infrastructure. For example, the University of Texas at Austin has established a standard battery of spoofing tests for GNSS receivers (TEXBAT), GNSS Vulnerability Test Systems are available from various vendors, Certification testing of GNSS receivers is available and required by some standards organizations, and the DHS (Department of Homeland Security) is establishing the Resilient PNT Conformance Framework to characterize all PNT sources and their behavior under cyberattack.
• Protecting the Individual PNT Source – GNSS receivers can be protected from cyberattacks using "smart antennas." These range from expensive, multi-element Control Radiation Pattern Antennas (CRPA), which track the individual satellites with narrow beams as they move across the sky and block interference, to inexpensive
  horizon blocking anti-jam antennas or simple two-element null steering antennas that search and block out interference.
• Monitoring and Logging Attack Events – the threat landscape is dynamic and we must adapt with new strategies. Creating threat libraries as they occur and sharing these across the PNT community helps to counter them.