2
Analysis Regarding the Three Study Tasks

2.1 TASK 1: APPROACHES TO EVALUATING HARMFUL INTERFERENCE CONCERNS

The following task is the first that has been put to this committee to consider:

Which of the two prevailing proposed approaches to evaluating harmful interference concerns—one based on a signal-to-noise interference protection criterion and the other based on a device-by-device measurement of the Global Positioning System (GPS) position error—most effectively mitigates risks of harmful interference with GPS services and U.S. Department of Defense (DoD) operations and activities.

In this task, the committee is charged with resolving a dispute between the GPS community and Ligado about which of the two methods is best for characterizing harmful interference. The committee’s response, discussed in this section, is that neither method is satisfactory in the simplistic forms in which they have been proposed.

2.1.1 Overview of Response to Task 1

As they were applied, neither of the prevailing approaches effectively mitigates the risk of harmful interference. As implemented, the signal-to-noise ratio (SNR) approach was too inflexible, and the position approach was too narrow in its applicability. Yet, both approaches do have a role in evaluating harmful interference to existing receivers. Of the two, the SNR-based approach is the more comprehensive and informative. By informing

the degradation in link margin,¹ this approach can be used to predict harmful impacts across a broad set of use cases: signal acquisition, real-time kinematic (RTK), timing, and so on. Neither approach provides an analytical, repeatable, or straightforward criteria to evaluate new entrants.

The determination of harmful interference is dependent on the particulars. There is an incredibly wide array of GPS use cases, including car navigation, network timing, precise surveying, and geophysical applications, to name only a few. These use cases have different failure modes, which result in varying interference tolerance. The most appropriate approach must be mapped to each use case. As such, the use of a single, firm SNR-based interference protection criterion (IPC) is not practical when applied to device-by-device performance. Although evaluation of device-by-device SNR degradation is informative and recommended, the criteria must incorporate the fact that harmful interference depends on use case. For example, some applications are harmed when code-lock is lost, while other applications are harmed from loss of carrier phase-lock.

The commonly advocated 1 dB SNR loss criteria has not been linked to the definition of Harmful Interference. Although a “1 dB criterion” precludes harmful interference in virtually all use cases, the vast majority of GPS use cases do not experience harmful interference at such a low level—that is, they have significantly higher link margin than 1 dB. As such, the 1 dB criterion is prophylactic, but conservative.

Task 1, as posed, does not directly address the bigger challenge: regardless of which approach is applied, drastically different conclusions can be reached. There are numerous test design particulars that must be considered, including determining the path-loss model, desired stand-off range, antenna coupling, degree of insensitivity of a particular receiver’s design to adjacent-band power, and performance thresholds. Even for a given use case, these issues are not easily resolved. Furthermore, a per-device SNR threshold creates a moral hazard—receiver manufacturers are incentivized to design adjacent-band-susceptible receivers in order to claim spectral easements.

Ultimately, both of the proposed approaches are cumbersome, owing to the intensive testing required. They do not provide an engineerable, predictable standard that new entrants can readily use to evaluate impact. As such, these approaches impede spectral progress. A new applicant for emissions in a new adjacent channel will have great difficulty in determining the emitter power levels and stand-off distances that will be guaranteed not to cause harmful interference to the existing GPS receiver base. A GPS receiver designer will be unable to design a receiver that will be guaranteed to be insensitive to unknown potential future allowed levels of adjacent-band power that might be allowed if the Federal Communications Commission (FCC) were to amend its Order

___________________

¹ “Link margin” refers to the difference between the observed post correlation SNR (in dB) and the minimum operating threshold (in dB).

20-48 to allow additional adjacent-band power beyond that contained in Box 1-1 in this report.

The next two sections discuss these two approaches in more detail. Their strengths and weaknesses are noted.

2.1.2 Considerations Regarding an SNR IPC

The SNR IPC approach consists of testing GPS receivers at various interference levels that could be attributed to Ligado transmissions and measuring the resulting degradation in reported C/N₀. C/N₀ is the standard industry-wide metric that is accepted by GPS subject matter experts to measure the operational health of a GPS receiver. Many key receiver functions can be mapped to C/N₀ thresholds—that is, C/N₀ level can be used to reliably predict a receiver’s ability to acquire signals in cold-, warm-, and hot-start. It can also be used to predict code- and carrier-lock. As such, it informs receiver performance over a wide set of operating conditions, including RTK. Insight into these core functions can be extrapolated to four GPS performance metrics: accuracy, continuity, integrity, and reliability.

To be clear, the measurand is the receiver reported C/N₀. This is computed internally by the GPS receiver, within its digital logic, typically measured from the correlator outputs, as described above. This is a device-estimated quantity, which is noteworthy because Task 1 implies that the C/N₀ criteria is a physical standard independent of receiver design and thus subject to purely analytical assessment, which is incorrect. Furthermore, because this measurement is made after the correlation, it blends the effects of both in-band and adjacent-band interference, with the latter being dependent on the receiver’s filtering design. In many regards, the intermingling is ideal because the receiver-reported C/N₀ lumps together all impacts from the interference. Nevertheless, it should be observed that there is not a deterministic mapping between receiver-reported degradation of effective C/N₀ and the physical power of the radio frequency (RF) wavefront as it arrives at the antenna, especially the adjacent-band power.

Generally, receiver-reported C/N₀ can reliably and consistently measure changes to better than 1 dB of accuracy. The key here is that the change in C/N₀ is being measured. There will be biases in the absolute C/N₀ estimate. That bias, however, falls out when looking at changes. The claim of “better than 1 dB accuracy” is a consequence of two effects: (1) the standard C/N₀ estimators are sufficiently linear over the region of interest (20–50 dB-Hz), and (2) the noise in the estimators can be effectively averaged down in reasonable bandwidths (averaging with less than a minute of data is sufficient). The accuracy of the delta C/N₀ measurement is only for a specific device. Across devices, there will be variation owing to implementation specifics (hardware and firmware), as previously discussed. There is a cause for caution because, as an internally generated

figure, there is no guarantee that every device’s receiver reported C/N₀ is computed using a widely accepted calculation. Additionally, real-world, live sky testing needs to be considered carefully because the carrier power can vary over time from multiple causes—satellite motion, antenna tilt, multipath, and so on. Simulator testing with controlled, or constant, signal power removes this concern.

As discussed earlier in this chapter and in the explanation of interference in Section 1.4, C/N₀ degradation is highly predictive of receiver performance, although identifying the cause of the degradation is less predictive and is receiver dependent. It can be used to inform impact across a range of regimes: cold-start, warm-start, code-tracking, and carrier-tracking. The required C/N₀ thresholds for these operational modes are known to provide reasonable accuracy for most receivers. As such, receiver-reported C/N₀ degradation provides insight into how a given GPS receiver’s performance will be impacted across a variety of use cases. Harmful interference must be tied to the available margin a receiver has for its intended task. Some applications, such as a car-based GPS, may depend only on code-lock. RTK and scientific applications typically depend on phase-lock and on mean time between cycle slips. A stationary receiver mounted adjacent to a Ligado transmitter may be limited by acquisition performance. The thresholds for these applications can vary significantly, by tens of decibels. Therefore, the same receiver model may be impacted in one use case but not another. As such, a single hard IPC threshold for C/N₀ is inadequate.

In practice, the question of how much degradation occurs to C/N₀ is highly receiver dependent. The Ligado emissions that fall directly into the GPS band can be considered separately from the emissions that fall outside the GPS band. The impact from Ligado emissions in the GPS band is not receiver dependent. This portion raises the noise floor and can be computed analytically. The Ligado signal components outside the GPS band affect C/N₀ in the GPS band via the GPS receiver RF front-end electronics (see Section 1.4.5). An important takeaway is that, depending on the receiver design, the levels of power that get translated from adjacent-band to GPS in-band can vary by 60 dB or more, as evidenced by Figure 2-1. In the Ligado downlink band from 1526–1536 MHz, there is a 60–65 dB variation of the emitted out-of-band power levels that result in a 1 dB reported loss of C/N₀ for various classes of receiver. It is reasonable to infer that these large differences in reported C/N₀ degradation stem from large differences in the various receiver classes’ ability to tolerate adjacent-band signal power. All of these receivers reject adjacent-band power to a sufficient degree to operate well in the pre-Ligado adjacent-band spectrum for which they were designed. Some, however, appear to have insufficient adjacent-band power attenuation once the approved Ligado downlink signals are present.

2.1.3 Considerations Regarding a Receiver Position Error IPC

At least for some use cases, position error is a reasonable metric in measuring harmful interference. As the primary metric of interest for many users, it is appropriate to consider accuracy. With that said, the applicability of this approach is too narrowly focused, and it will not inform harmful interference over the entire set of use cases—for example, this approach does not directly address the three other metrics of interest: availability, continuity, or integrity.

It is not possible to assess harmful interference across a range of use cases using only position error as the observable. In order to do such an assessment, testing would need to include performance assessments across GPS operational states, including cold-start and warm-start acquisition and phase-lock. Such testing would be laborious. Furthermore, the evaluation of position error must carefully consider the behavior of the receiver. The dynamic model and the method and manner of the phase measurement fusion can greatly impact the position accuracy. RTK and other differential approaches must be evaluated relative to their expected performance. Harmful interference in such

use cases could be only centimeters of extra error. Additionally, some applications may experience harmful interference in manners that are not at all observable using position error—for example, applications interested in timing or velocity or scientific applications that are primarily interested in raw phase measurements. Long, perhaps prohibitive, testing would be necessary to develop statistics on mean time between cycle slips, or time-to-first-fix. These drawbacks are mitigated with SNR-based approaches.

As with the SNR-based approach, the application of uniform hard thresholds for accuracy do not make sense. Harmful impact to accuracy depends on the use case, and it should be mapped as such. The Roberson report applied an ambiguous approach to determining harmful impact based on position error.

The implementation of the position error metric also requires careful consideration. The Roberson report used a 3-minute time average to compute position error. This is not at all unreasonable when applied to some use cases. However, it would not reveal harmful interference in all use cases. To provide an example, interference might not affect average position error very much, but it might affect continuity of operations, which can be highly problematic for aviation devices or high-precision devices. Losing signals even for a very short time during a surveying operation or at an Airport Ground-Based Augmentation System receiver could cause immense operational problems for the user. If some of the lowest elevation satellites become unusable, then average position error might not increase much under standard testable conditions, but that loss might make a receiver’s defenses against a spoofing attack much less robust. Therefore, simple measures of position accuracy degradation do not adequately capture all of the possible harmful interference effects that are of legitimate concern. Even if position accuracy degradation were the accepted metric of whether harmful interference had occurred, the Roberson study was inadequate for purposes of making that assessment. The testing that was completed was limited to only a subset of GPS use cases. For example, it did not consider high-bandwidth error effects. In the high-bandwidth case, error standard deviation is also important. The question of error standard deviation versus error mean is further addressed near the end of Section 2.2.4.

2.1.4 Comparison Between SNR and Accuracy Approaches

All things considered, the two approaches have more in common than not. Table 2-1 compares the approaches across a number of considerations. Both approaches are inherently device-by-device measures of harmful interference that provide little insight into the physical mechanism within the receiver that caused the degradation. Besides the effort required to conduct the device-by-device testing, such testing opens questions such as the set of devices to test, and where to draw the line on the acceptable level of harmful interference. Is the interference considered harmful if any of the tested devices

TABLE 2-1 Comparing Approaches

NOTE: Color coding scheme: red highlights undesirable metric properties; green highlights desirable properties; and yellow highlights lie between being desirable and undesirable.

experiences harm, or is the criterion that the “median” receiver must experience harm, or is the criterion that all receivers must experience harm? Even if all parties agree on which of the two approaches to use, considerable disagreements are likely to remain with the implementation of the approach. Decisions regarding the proper path loss model, or the acceptable impact region size, or which thresholds to use, or which receivers should be guaranteed to maintain acceptable performance would still need to be settled. How much SNR margin do receivers need? What is an acceptable frequency for cycle slips? The list goes on, and only through dialog among interested parties can these multitude of concerns be addressed.

The primary distinction between the two approaches is their predictive power. The SNR-based approach informs link margin. The link margin can forecast performance across most, if not all, of the receivers’ operating modes. This simplifies testing to a large degree, but it does not make the testing simple. Moreover, with the SNR approach not only are pass/fail criteria known, the remaining margin is also known. The limitations of the SNR-based approach contemplated in Task 1are the implicit usage of a single, firm

IPC threshold and the implicit dependence on receiver RF front-end design. Harmful interference occurs at a wide range of SNR degradation levels depending on receiver and use case, and a wide range of SNR degradation levels occurs in different receivers for the same Ligado transmission power. The receiver position error method of evaluation, on the other hand, has little predictive power under any circumstances beyond the specific use cases and receivers covered in the tests. Had the question been posed in a way that allowed for the nuances of the SNR approach and with some recognition of the applicability of receiver standards, the committee likely would have chosen SNR degradation as the clear favorite for deciding whether or not Harmful Interference had occurred.

2.2 TASK 2: HARMFUL INTERFERENCE TO GPS AND MOBILE SATELLITE SERVICES

This section considers the second task of the committee:

2.2.1 Overview

The evaluation of harmful interference on services using the MSS/GPS band has been thoroughly investigated. The conclusions presented here are based on the available test data, analysis, and reports.

For GPS, several sets of interference tests have been performed that span many representative GPS devices drawn from many different receiver classes and suppliers. The tests evaluated various scenarios and advocated for different metrics to determine the onset of harmful interference. Despite these differences, the results consistently indicate that a majority of the devices tested do not experience harmful interference.

Based on the results of tests conducted to inform the Ligado proceeding, most commercially produced general navigation, timing, cellular, or certified aviation GPS receivers will not experience significant harmful interference from Ligado emissions as authorized by the FCC. High-precision (HP) receivers are the most vulnerable receiver class, with the largest proportion of units tested that experience significant harmful interference from Ligado operations as authorized by the FCC.

The committee found that it is within the state-of-the-practice of current technology to build a receiver that is robust to Ligado signals for any GPS application and that all GPS receiver manufacturers could field new designs that could co-exist with the

permitted Ligado signals and achieve good performance even if their existing designs cannot.

For mobile satellite services (MSS), Globalstar is unlikely to experience harmful interference (only the uplink is in the L-Band, and it is a code-division multiple access [CDMA] signal). Iridium terminals will experience harmful interference on their downlink caused by Ligado user terminals operating in the UL1 band within a significant range of a Ligado emitter—up to 732 m. (See Section 2.2.5.)

2.2.2 Impact

The difficulty of determining a metric to reliably quantify the impact of interference is owing in part to GPSs being used in different ways by the communities that need higher precision. The most common use case is a roving user with modest accuracy needs. Such a user will observe interference intermittently, if at all. The most severe harmful impact is the persistent loss of code-lock tracking on multiple satellites, although this will be very rare. Yet, many users will be impacted in other ways. Stationary users may be harmed by loss of signal acquisition. High-precision users depend on stable phase-lock. Each of these use cases is impacted at significantly different interference levels.

As licensed, some aviation receivers may experience harmful interference. The conditions of FCC Order 20-48 stipulate that Ligado cannot operate a downlink (D/L) antenna at “any location less than 250 feet laterally or less than 30 feet below an obstacle clearance surface established by the Federal Aviation Administration (FAA).” Certified aviation receivers will not experience harmful interference outside of this exclusion zone, and fixed-wing manned aircraft operating under standard flight rules will never enter these cylinders of exclusion.² Other aviation users, however, cannot be guaranteed to remain outside these cylinders. Public safety helicopter operations, which are conducted by medevac providers, law enforcement agencies, firefighting departments, the U.S. Coast Guard, the Army National Guard, and other entities, often work in close proximity to infrastructure during the course of their operations. Crews do their very best to mitigate risk and ensure safety through a variety of methods, including visual separation and reliance on important installed safety systems, such as GPS and radio altimeters. After Ligado deployments, helicopter operations will continue to operate close to towers and other infrastructure as required by the specific mission or operation. While most such activities will operate under Visual Flight Rules within 250 feet of a Ligado tower, no guarantees can be made for GPS-only navigation within the 250 × 30 feet cylinders.

___________________

² E. Drocella, C.-W. Wang, and N. LaSorte, 2020, “Assessment of Compatibility Between Global Positioning System Receivers and Adjacent Band Base Station and User Equipment Transmitters,” NTIA-TM-20-536, Section 5.2.3.9, U.S. Department of Commerce.

The same can be said for GPS-only navigation by unmanned air vehicle systems that will provide critical services in the future. Such systems may experience harmful interference to their GPS navigation subsystems when inside the cylinders and must rely on other sensor input, including cameras. Ligado interference may occur exactly when GPS is informing a critical safety service, such as GPS-based terrain avoidance after an unexpected transition to low visibility conditions. Supplementary navigation instruments, including radar and fixed beaconing, may mitigate the occasions when visual flight rules cannot be relied on and proximity to a Ligado tower results in loss of GPS, but there is presently no guarantee of this.

The accuracy of high-precision applications is complicated and cannot be reduced to a simple assessment of the impact of carrier-to-noise density ratio, or the impact of a 1 dB reduction in that ratio.

2.2.3 Test Results

To assess interference with adjacent channel Ligado signals on GPS performance, key tests were performed by four different groups:

• Roberson and Associates (RAA), a consulting firm sponsored by Ligado;
• National Advanced Spectrum and Communications Test Network (NASCTN), a confederation of the federal spectrum and testing experts, commissioned by Ligado;
• U.S. Department of Transportation (DOT), which created the Adjacent Band Compatibility (ABC) report; and
• A DoD classified assessment that was viewed by a cleared subset of the committee.

The DOT ABC assessment included two primary components, one led by the DOT Office of the Assistant Secretary for Research and Technology and focused on civilian GPS devices and their applications, and the other led by the FAA and focused on certified GPS avionics.

Appreciating the Difficulty of Testing

These testing studies came up with varying and conflicting assessments of interference potential, which is not unreasonable given the complexity of the assumptions, differences in metrics, and experimental processes. Each testing study has its relative strengths and weaknesses, and they are outlined below, along with key results. Testing is an essential part of the regulatory assessment; it does help in reducing, but not eliminating, uncertainty regarding underlying assumptions or circumventing the need for all the details.

Testing has its limitations. Not all makes and models of GPS receivers produced can be tested and a representative set is needed. Those GPS receivers selected must be tested under fair, realistic conditions. Metrics depend on what needs to be tested and how that metric relates to harmful interference. The FCC regards harmful interference as interference that impacts the functionality of a device; again, functionality is a subjective measure. In the case of GPS, functionality is the performance of the GPS device, which is somewhat subjective and use-case dependent. Thus, identifying a single metric to characterize Harmful Interference is the most challenging part of creating a non-controversial testing process.

Although all tests looked at the C/I (or C/N₀) metric, this metric is self-reported from the device, and the measurement is non-standardized in concept and in hardware implementation, making it a problematic metric. (See Section 2.1.) Variations of up to 62 dB in device-reported C/I were measured between devices under the same conditions in the ABC study.

Furthermore, applying the test results to a relevant operational scenario requires the selection of a propagation model to determine the range at which the observed effects will occur. These models are highly dependent on the geometry of the scenario and can vary by many tens of decibels. A conservative approach uses free space propagation loss which is useful for cases where there is a clear line of sight (LOS) between the transmitter and the receiver. A non-conservative approach is to use a non-LOS model, which would occur if the receiver were masked from the transmitter by buildings or terrain. Both of these approaches as well as approaches that combine the two (LOS for distances < ~100 m and non-LOS for longer distances with a smooth transition between) were used in the various reports.

The committee realizes that all testing is subject to limitations, but that testing can be used to evaluate the possibility of harmful interference if appropriate post-test analyses are applied. One obvious analysis is to determine equivalent stand-off distances under a given propagation model. It would be impractical to test all aspects of all significant use cases and modes of harm to receiver operations, but it is possible, in theory, to deduce when the various modes of harm will occur in the various use cases based on receiver test data of a more limited nature. In order to enable such analyses, it is necessary to provide the raw (post calibration) data from all of the tests;³ only the NASCTN “LTE Impacts on GPS” tests did this. In addition, no guidance was given about which receivers should be protected from harm. Should it be all receivers tested, only those with a certain market penetration, only those serving highly important applications, or only those with the best RF front ends? Therefore, despite all the tests, the available data

___________________

³ Dependent variables were self-reported by the devices under test, which are dependent on the receiver point design.

and analyses fell short of what the committee would have liked to see in order to make the best possible decision about whether the permitted Ligado tower transmission levels would cause unacceptable harm to GPS receivers. Nevertheless, the committee was able to draw some conclusions about harmful interference based on the test data, on rudimentary analyses of its own, and on information provided by various entities that gave presentations to the committee.

It is important to note that no mobile satellite systems interference tests were presented to show the impact on systems such as Iridium. Iridium presented an analysis. Several members of the committee produced their own independent calculations that verified the correctness of the core finding of the Iridium analysis: that interference to its downlink signals will occur. DoD provided a classified test report from 2016 to quantify the impact of Ligado interference on their systems.

Overview of Test Approaches and Results

The test objectives, setup, limitations, and relevance to FCC Order 20-48 are summarized in Table 2-2.

Both the NASCTN and RAA reports did show that some high-precision receivers could be adversely impacted by Ligado emissions. The RAA study showed that high-precision receivers it tested could be fixed by replacing the antenna with one that is more frequency selective (when replacing an external antenna is possible), but the testing did not include HP functionality, such as RTK. For some platforms, such as weapon systems or aircraft, replacing an antenna may not be feasible. Military systems often leverage the L2 GPS signal, which will make it much more robust to interference if the additional L2 reception has been incorporated primarily as a means of increasing the receiver’s robustness to jamming. If, on the other hand, a particular application requires both the L1 and L2 signals in order to remove the ionosphere error term, then the use of L2 will not mitigate any harmful interference to the L1 GPS signal that might be caused by Ligado.

2.2.4 GPS

The radionavigation satellite services (RNSS) band ranges from 1559–1610 MHz and contains the GPS L1 band. There are several sources of interference that can fall into this band, as illustrated in Table 2-3.

GPS In-Band Interference

Noise levels in the GPS band are approximately −174 dBm/Hz, so at a distance of 10 meters from a Ligado transmitter limited to −130 dBm/Hz in the GPS band (56.5 dB free space path loss at 1584.5 MHz), the interference from the Ligado user or base station will be at least 12 dB below the noise floor.

TABLE 2-2 Tests Relevant to FCC Order 20-48

TABLE 2-3 Sources of Interference in GPS Band

For the case of multiple base stations, the separation requirement of 433 m ensures that a given GPS device is likely only to be impacted by a single Ligado base station. For the multiple user case, four users would need to be within a 10 m radius of the GPS receiver and be transmitting simultaneously to reduce the GPS C/N₀ by the 1 dB advocated for by the GPS vendors.

As a result, the in-band interference allowed by FCC Order 20-48 users or base stations are unlikely to have any adverse effect on GPS operations.

GPS Out-of-Band Interference

Out-of-band emissions (OOBE) near GPS (less than 1559 MHz or greater than 1610 MHz) can potentially impact GPS receiver performance. The degree of interference will depend on several receiver design factors, including the receiver front-end bandwidth and dynamic range, the sampling rate, and signal processing used. The power, frequency, directionality, and duty cycle of the interferer are also critical parameters.

In existing GPS receiver designs, aliasing of out-of-band signals into the GPS band is typically not a problem. However, the large power discrepancies between the Ligado base station/user and the GPS signal make it possible for aliased power to be folded into the GPS band and may adversely affect receivers not designed for that environment. The center of the RNSS band is at 1584.5 MHz, the Ligado base station downlink is at 1531 MHz, the lower Ligado user uplink is at centered 1632.5, and the upper Ligado user uplink is at 1651.5 MHz. The frequency difference between the center frequencies of the Ligado signals and the RNSS band are 53.5 MHz, 48 MHz, and 67 MHz for the Ligado downlink, lower uplink, and upper uplink respectively. A GPS receiver with a sampling rate below twice the difference frequency must employ adequate filtering before the A/D converter to avoid aliased interference. This design constraint is normal for any digital system. As an example, a GPS receiver at a distance of 10 m (56.5 dB path loss) from a

base station (39.8 dBm), would require an attenuation of 111 dB to be at the same level as the GPS signal (−128 dBm), which would need to be provided by the analog filtering in the front end.

Note that the exact frequency plan (intermediate frequency [IF], sampling rate, number of down-conversion stages) for a particular GPS receiver is a critical design parameter, and receiver design depends heavily on the spectral environment and the use of the receiver. GPS receivers operating in a quiet environment will require less attenuation, while receivers that use corrections delivered via an MSS signal will need to receive those frequencies as well. The current Ligado downlink band (1526–1536) was used to deliver these corrections, but they have moved several times up to 2016, when Ligado agreed to permanently locate them in the 1555–1559 MHz range.⁴ As such, the Ligado downlink band is within the passband of some legacy receivers.

In summary, dealing with OOBE is a standard design problem for digital receivers and can be accommodated through a number of well-understood design techniques as long as the environment is well characterized and understood. The committee found that it is within the state-of-the-practice of current technology to build a receiver that is robust to Ligado signals for any GPS application, and several manufacturers have done so. The GPS/MSS band environment has been in constant flux since the introduction of the Ancillary Terrestrial Component (ATC) rule, making it difficult if not impossible for receiver vendors to anticipate the environment a receiver will encounter over its lifetime. Furthermore, every rule change risks making obsolete equipment with design assumptions that were violated by the change.

Test Results

C/N₀

Several tests were designed to evaluate the degradation in C/N₀ that occurs as a result of an interfering signal. The plots in the figures below derive from National Telecommunications and Information Administration (NTIA) TM-20-536, which analyzed data from several previous test events to inform the community regarding the proposal that was being considered in the 2020 time frame (and largely manifested itself in FCC Order 20-48). The data constitute an aggregate from tests performed by the DOT Adjacent Band Compatibility, DoD, Roberson and Associates, National Advanced Spectrum and Communications Test Network, 2011 FCC Technical Working Group, and 2012 National Positioning, Navigation, and Timing Systems Engineering Forum. The data look across a large population of receivers in several different categories, high precision (HP), general

___________________

⁴ S. Riley, Trimble, Inc., 2021, “Trimble Presentation to NAS,” Presentation to the Committee to Review FCC Order 20-48, January 12, 2022, Washington, DC: National Academies of Sciences, Engineering, and Medicine.

location/navigation (GLN), timing (TIM), cellular (CEL), non-certified general aviation (GAV), and space based (SPB) that were evaluated with varying scenarios in the different studies.

Plots showing the percentages of receivers that experienced 1 dB, 3 dB, and 5 dB degradation in C/N₀ at each interferer power level are shown for the HP, GLN, TIM, and GAV classes for a single base station (DL) or a single user handset (UL1/UL2). In addition, vertical range lines indicate the levels that would be experienced by a receiver at 1 m, 10 m, and 100 m from the interferer using a standard free space loss model, 20 log₁₀ (4πd/λ) at the center of the GPS band. These reference lines do not include antenna gain/polarization losses (typically ~5 dB), which are situation dependent and can be accounted for by translating the lines to the left. Note that the number of receivers in each case varies based on the tests that were performed in each of the campaigns from which those data were aggregated. For uplink cases, no power control was assumed, which is indicative of the case where the user terminal is on the periphery of the coverage area. This is conservative, but the area in a coverage region is concentrated in the regions farthest from the base station. The impact of power control for a particular user can be considered by scaling the interferer power axis for a given range to the base station.

Ligado Base Station Interferer

Figure 2-2 is for a single Ligado DL interferer in the 1525–1535 MHz or 1526–1536 MHz band, depending on the study the interferers were drawn from. The range lines use a 9.8 dBW base station power per FCC Order 20-48.

Ligado UL1 Interferer

Figure 2-3 is for a single Ligado UL interferer in the 1627.5–1637.5 MHz band. The range lines use a −7 dBW user power per FCC Order 20-48.

Ligado UL2 Interferer

Figure 2-4 is for a single Ligado UL interferer in the 1646.5–1656.5 MHz band. The range lines use a −7 dBW user power per FCC Order 20-48.

In looking through the cumulative distribution plots, many GPS receivers that were tested do not show a significant degradation in C/N₀ even at very close ranges. There are, however, many GPS receivers that show 1–5 dB degradation even at ranges beyond 100 m. FCC Order 20-48 allows a deployment with 433 m between base stations or a maximal range of 250 m to a base station. On the uplink side, mobile Ligado users can be

numerous, and can be anywhere in the coverage area, each contributing to interference; however, interference will be dominated by the closest user(s), owing to the path loss.

Mean and Standard Deviation of Position Error

In addition to C/N₀, several studies have evaluated the use of the position error as a metric to determine harmful interference. The RAA study evaluated 2D position error, and the NASCTN study evaluated 3D position error. Figure 2-5 shows the mean and mean +3 standard deviations of the data overlaid on a scatter plot of the 3D position error versus the

interferer power level at the GPS receiver using data from the NASCTN study. In addition, vertical lines indicate power levels that correspond to ranges between the receiver and the interferer of 100 m, 10 m, and 1 m as calculated using free space path loss. As an example, a GLN device (DUT2) is shown on the left. In this particular case, the spread of the position error, the mean, and the standard deviation of the position error are relatively stable with respect to the interference power level. The figure on the right shows a similar plot for a high-pressure processing (HPP) device (DUT7). In this case, the mean remains relatively flat, but the spread increases by more than a factor of 4 at a power level that is consistent with a range of greater than 100 m, indicating degradation owing to interference.

The NTIA TM-20-536 report looks at similar parameters from the NASCTN and research and analysis studies and additionally correlates the position error with receiver-reported C/N₀. The conclusion drawn from the aggregate of all of the data is that the mean of the position error does not indicate the increased spread in the positioning error whereas the standard deviation provides a more suitable indicator of the impact of the interference on the quality of the data provided by the device. Therefore, the mean of the position error should be augmented with additional error statistics (standard deviation, root mean square [RMS], sample error distribution, etc.) in the evaluation of harmful interference to capture the increased spread in position error with increased interference.

Similar plots for UL1 and UL2 for these same devices appear in Figure 2-6.

2.2.5 Mobile Satellite Services

The MSS band supports uplinks and downlinks between mobile users and satellite relays. (See Figure 1-2.) The band, discussed above in Section 1.1.2, is currently divided into three sections: GEO downlinks (1525–1559 MHz), GEO uplinks (1626.5–1660.5 MHz), and Big LEO uplinks/downlinks (1610–1626.5 MHz). GEO Earth terminals typically use high-gain antennas to compensate for the long range and thus have natural attenuation of signals that are not line-of-sight to the satellite. The focus here is therefore on the impact of ATC OOBE on the Globalstar and Iridium systems.

Earth to Satellite

Interference on the satellite uplinks is mitigated by limiting Ligado OOBE into the uplink bands and through the radiation pattern, which directs energy toward the terrestrial users instead of into space.

The Globalstar system’s uplink transmissions are at 1610–1618.725 MHz, and use a CDMA waveform that inherently provides some immunity to interference.⁵ The downlink is in the C-Band. The frequency plan of the Globalstar system coupled with the levels of the ATC OOBE (−130 dBm/Hz at 1610 MHz to −101 dBm/Hz at 1618.725) into the Globalstar uplink band do not result in interference at the satellite from the ATC emissions.

___________________

⁵ Globalstar, Inc., 2017, “Globalstar Overview,” https://www.globalstar.com/Globalstar/media/Globalstar/Downloads/Spectrum/GlobalstarOverviewPresentation.pdf.

The Iridium system uses a frequency-division multiple access/time-division multiple access multiplexing scheme, so the uplinks and the downlinks share the frequencies from 1617.775–1626.5 Mhz. The ATC OOBE into the Iridium band are –104.7 dBm/Hz at 1617.775 MHz to −76.26 dBm/Hz at 1626.5 MHz. Similar arguments can be made for the Iridium uplink as for the Globalstar uplink and despite the higher interference power levels at the satellite, no claim of interference has been made.

Satellite to Earth

The Iridium downlink shares the same frequencies as the uplink and is subject to the same interference. Iridium has stated their noise floor is −170 dBm/Hz, which is

consistent with a noise-limited system.⁶ Interference levels that approach this level will have an impact on Iridium performance. Unlike the uplink case, the radiation pattern will direct the energy toward the Iridium terminal.

The OOBE spectral mask for ATC users is a linear ramp from −73 dBm/Hz at 1627.5 MHz to −130 dBm/Hz at 1610 MHz followed by a linear ramp to −135 dBm/Hz at 1608 MHz, as shown in Figure 2-7. In addition, the OOBE from the upper uplink band is −64 dBm/Hz at 1645.5 MHz with no clear lower bound until the −73 dBm/Hz restriction at 1627.5 MHz. When multiple users are present, the spectral masks from each user will be additive after being adjusted for the range of each user to the Iridium receiver.

For the Iridium band, channels on the high side will see an interference level of −76 dBm/Hz from a single user, which will require 94 dB of attenuation to be reduced to the noise floor. This will occur at a distance of 732 m (free space path loss) or 51 m (non-line-of-sight).⁷ On the lower end of the band, the interference level is −102 dBm/Hz, which will require 68 dB of attenuation to be reduced to the noise floor. This will occur at a distance of 40 m (free space path loss) or 16 m (non-line-of-sight).

The committee’s conclusion is that the Iridium downlink will experience harmful interference at relevant ranges. The above analysis assumes a single user; the situation will be both more likely and more severe as the spatial density of the users increases.

___________________

⁶ Iridium Communications, Inc., 2020, “Petition for Reconsideration,” FCC Proceedings IB 11-109 and IB 12-340, Table 3, May 22.

⁷ Non-LoS range determined using the COST 231 model for an urban environment with a frequency of 1626.5 MHz, Tx antenna height of 4 m, Rx antenna height of 1.5 m, average roof height of 15 m, distance between buildings of 35 m, street width of 17.5 m, and a 45-degree propagation angle.

2.2.6 U.S. Department of Defense

DoD uses the MSS band for many different applications, including communications, navigation, tracking of military assets such as equipment and vehicles (including drones), synchronization of timing, and so on. In addition, the U.S. government is by far the single largest Iridium customer.

GPS can be relatively easy to jam; numerous studies have indicated this. One can even find commercially available but illegal GPS jammers on the market that are used to defeat tracking systems. Military system GPS receivers are more difficult to jam. The military uses an augmented L2 signal that provides resilience to jamming around L1, if it is not needed for ionospheric corrections. (See Section 2.2.3.) Other techniques such as null-steering antennas can provide additional robustness.

DoD is also a major customer of commercial mobile satellite systems, such as Iridium. DoD has evaluated the impact of FCC Order 20-48 on department devices and missions. The following summary points were provided to the committee in a set of slides dated March 15, 2022. It is important to note that these conclusions were asserted by DoD without providing publicly available supporting data and were not discussed by the committee in a public session.

• DoD and interagency partners conducted testing to determine the impacts to GPS (captures FCC Order 20-48’s authorized deployment). The tests demonstrated that the proposed signal introduces harmful interference to critical national security mission capabilities.
• The terrestrial network authorized by FCC Order 20-48 will create unacceptable harmful interference for DoD missions. The mitigation techniques and other regulatory provision in FCC Order 20-48 are insufficient to protect national security missions.

In a subsequent response to questions from the committee, DoD also commented that there were no independent studies or technical analysis of Iridium’s claims of interference, and that although DoD did initiate its own analysis of some of the claims, that work was not completed as essential information was unavailable to DoD.⁸

Additional information on the test results and analysis as they related to DoD systems and missions is discussed in a classified annex to this report.

___________________

⁸ DoD responses to committee questions, April 29, 2022.

2.3 TASK 3: FEASIBILITY, PRACTICALITY, AND EFFECTIVENESS OF MITIGATION MEASURES IN THE FCC ORDER

Task 3 concerns the feasibility, practicality, and effectiveness of the mitigation measures proposed in the FCC Order 20-48 with respect to DoD devices, operations, and activities.

2.3.1 Overview

The FCC Order 20-48 enumerates several potential mitigations when a receiver experiences harmful interference. These include:

1. Mandating exclusion zones for Ligado emitters (as was done for potential navigation receivers in the order).
2. Replacing components of vulnerable receivers (e.g., replacing antennas in some high-precision receivers was discussed in several reports).
3. Replacing all vulnerable GPS receivers.
4. Moving receivers farther away from Ligado emitters when practical.
5. Putting a regime in place to require Ligado to have a facility to turn off emitters in some geographic locations on notice.
6. Negotiated mitigations between Ligado and the affected government agency to determine “an acceptable received power level over the military installation.”

In addition, the FCC requires Ligado to maintain a database of Ligado emitters. The database will be helpful in ruling out Ligado’s influence in harmful interference in many cases and help identify potentially interfering Ligado emitters, but attributing harmful interference as experienced by identified receivers to one or more Ligado emitters will be difficult.

In the case of harmful interference with GPSs, the effectiveness and practicality of any of the foregoing potential mitigations depends on the type of receiver and the application. It is important to distinguish between two types of equipment:

1. DoD Authorized/Compliant Devices approved for weapons and weapons delivery systems and other national security certified devices.
2. Commercial GPS Devices when used in national security applications with an express waiver or navigational warfare (NAVWAR) compliance determination per DoD Instruction 4650.08 and Chairman of the Joint Chiefs of Staff Instruction (CJCSI) 6130.01 or commercial devices that are used in other DoD operations or missions such as emergency response or partner operations.

In its response to Task 2, the committee assesses that harmful interference with commercial devices is substantially unlikely with respect to CEL, GLN, and TIM devices as well as HP devices sold after circa 2012.⁹ The committee also believes that it is reasonable to assume that DoD Authorized/Compliant Devices are expected to withstand willful interference under substantially higher power than is authorized under FCC Order 20-48. However, no specific information in this regard was presented or made available as part of the committee’s public study.

Notwithstanding this, the analysis below evaluates the potential mitigations in the context of device type and application below.

In addition, aside from GPS applications, DoD operations may employ Iridium services that may be affected by Ligado uplink emissions near 1627.5 MHz, as is discussed below. Only potential mitigation option 6 above is applicable in this case, and the committee does not have enough information to assess its potential effectiveness or practicality.

2.3.2 Inside the United States

As noted above, the committee assumes that DoD Authorized/Compliant GPS receivers or systems that incorporate such receivers used inside the United States are built to withstand jammers emitting much higher interference power than the authorized Ligado emissions, jammers that are purposely designed to interfere with civilian and military GPS in a NAVWAR environment. While such DoD Authorized/Compliant GPS receivers are unlikely to experience degradation owing to Ligado emissions, the committee did not have access to concrete information about the actual prospect of harmful interference in these devices.

If any DoD Authorized/Compliant Devices GPS receivers or systems that incorporate such GPS receivers used inside the United States experience harmful interference, there are relatively few satisfactory mitigations because these systems must pass very long and expensive operational test certification; generally, actions that include replacing antennas or electronics to provide mitigation would involve unsatisfactorily long delays. In some cases, replacing older devices with newer versions of such devices that are protected from harmful interference and already qualified may provide a plausible solution. However, as noted above, the likelihood of such interference, especially given L2 resilience, seems very low.

The remainder of this section’s comments apply to DoD’s use of Commercial Devices. First, potential harmful interference to GPS receivers from Ligado emitters in

___________________

⁹ See, for example, S. Riley, Trimble, Inc., 2021, “Trimble Presentation to NAS,” Presentation to the Committee to Review FCC Order 20-48, January 12, 2022, Washington, DC: National Academies of Sciences, Engineering, and Medicine.

the 1526–1536 MHz band or GNSS band are considered, recalling that the committee judges the likelihood of such interference as low.

Replacing antenna subsystems in older commercial devices that experience harmful interference when close to a Ligado emitter can be effective. Not all vulnerable HP devices can practically be retrofitted in this way.

As pointed out earlier, if devices that may experience harmful interference can be moved farther away from a Ligado emitter, any harmful interference should be mitigated. For example, if such a receiver can be moved from operating within 3 meters of a Ligado emitter to 30 meters from a Ligado emitter, it will then experience a 20 dB decrease in interference power. Whether this is practical is application specific.

Replacing an older commercial GPS receiver with a newer one should be an effective mitigation. However, there is a cost (high in the case of HP receivers), and in some cases it may be difficult to actually substitute the devices.

When DoD employs Commercial Devices in operations that occur in fixed locations (such as a base), simply requiring that there be no Ligado emitters within 500 meters of the location should be highly effective and practical.

Last, in some cases, Commercial Devices may be replaced with alternative geolocation services, but this is likely to be expensive.

The principal mechanism identified by the FCC for mitigation requires joint exploration between Ligado and the entity experiencing a specific interference problem, and this mitigation is case specific. Additionally, attribution of GPS performance degradation to Ligado is potentially difficult to prove. When asked by the committee if DoD could describe any engagement with Ligado to resolve any specific or general interference scenarios or cases since the April 22, 2020, Order and Authorization, DoD responded, “Per NTIA direction there has been no engagement with Ligado since the April 22, 2020, Order and Authorization pending resolution of the petition for reconsideration.”

The previous discussion addressed the potential harmful interference from Ligado emissions on DoD operations as it relates to GPS receivers. This analysis now turns to potential harmful interference from Ligado’s uplink emissions in the band between 1627.5 and 1637.5 MHz. Uplink transmissions present no material impact to GPS service in the 1559–1610 MHz band.

A particular concern is for DoD’s use of commercial Iridium services just below 1626.5 MHz. Ligado uplink emitters near 1627.5 MHz may adversely affect Iridium terminals within up to about 750 meters of such a Ligado uplink emitter. The experimental record with respect to these effects is not nearly as exhaustive as for the devices in the RNSS band, but crude calculations indicate that harmful interference is possible. These calculations are based solely on the interference power permitted by the FCC order near 1627.5 MHz. In addition to communications services, Iridium provides supplemental

aviation navigation services. These should not be affected by emission from the Ligado system at altitude or when operating outside the United States.

The committee notes that there are conflicting interpretations of the regulatory requirements that apply to this potential scenario. In paragraph 117 of its order in the Ligado proceeding, the FCC, referencing a 2005 commission ruling,¹⁰ requires services in the 1626.5 MHz band to tolerate interference substantially higher than that which would be produced by a Ligado uplink transmitter. Iridium has countered^11,12 that FCC regulations place the onus on the interferer to remedy problems caused by the permitted level. Consistent with its intent to focus on whether there is or is not harmful interference, the committee makes no judgment about the regulatory questions raised here.¹³

There are several potential mitigation options for this interference. First, it may be possible to require much sharper filtering by Ligado uplink transmitters operating near 1627.5 MHz. Second, it may be possible to bar Ligado uplink emitters within, say, 1,000 meters of potentially affected DoD operations inside the United States. The practicality of this second mitigation requires that DoD have an accurate survey of such operations and locations and does not account for DoD operations outside fixed U.S. installations—for example, during a disaster assistance mission. Harmful interference may also be mitigated by a more interference tolerant design of Iridium terminals used by DoD. More interference tolerant designs may rely on coding gains, current Iridium high-bandwidth features may make this difficult and ultimately infeasible. As a practical matter, absent significantly more information on Iridium receiver design and Ligado transmitter characteristics near 1627.5 MHz accompanied by actual laboratory and field testing, it is impossible to predict the effectiveness of these technical mitigations. As a result, addressing this issue is most likely to require cooperative discussions among DoD, Iridium, and Ligado.

2.3.2 Outside the United States

The Ligado operation is licensed only in the United States, so there should be substantially no impact from the Ligado system to any DoD devices or operations outside the United States.

___________________

¹⁰ 47 CFR § 25.253(g)(1).

¹¹ See “Petition for Reconsideration from Iridium Communications Inc., Aireon LLC, Flyht Aerospace Solutions Ltd., and Skytrac Systems Ltd.” in the Ligado proceeding, document ID 10522231459169, received May 22, 2020, and posted May 26, 2020, p. ii, https://www.fcc.gov/ecfs/search/search-filings/filing/10522231459169, copy provided to the committee by Iridium on September 2, 2021.

¹² Letter submitted to the committee on February 22, 2022, which is available on request from the National Academies’ Public Access Records Office.

¹³ This paragraph was revised after this report’s initial public release in September 2022 to clarify that certain assertions were made by stakeholders or participants in the Ligado proceeding and that those assertions were not being made by the study committee.