By Christopher J. Hegarty
Based upon lessons learned from the LightSquared situation, the author identifies important considerations for GPS spectrum interference standards, recommended by the PNT EXCOM for future commercial proposals in bands adjacent to the RNSS band to avoid interference to GNSS.
On January 13, 2012, the U.S. National Positioning, Navigation, and Timing Executive Committee (PNT EXCOM) met in Washington, D.C., to discuss the latest round of testing of the radiofrequency compatibility between GPS and a terrestrial mobile broadband network proposed by LightSquared. The proposed network included base stations transmitting in the 1525 – 1559 MHz band and handsets transmitting in the 1626.5 – 1660.5 MHz band. These bands are adjacent to the 1559 – 1610 MHz radionavigation satellite service (RNSS) band used by GPS and other satellite navigation systems. Based upon the test results, the EXCOM unanimously concluded that “both LightSquared’s original and modified plans for its proposed mobile network would cause harmful interference to many GPS receivers,” and that further “there appear to be no practical solutions or mitigations” to allow the network to operate in the near-term without resulting in significant interference.
The LightSquared outcome was a lose-lose in the sense that billions were spent by the investors in LightSquared and, as noted by the EXCOM, “substantial federal resources have been expended and diverted from other programs in testing and analyzing LightSquared’s proposals.” To avoid a similar situation in the future, the EXCOM proposed the development of “GPS Spectrum interference standards that will help inform future proposals for non-space, commercial uses in the bands adjacent to the GPS signals and ensure that any such proposals are implemented without affecting existing and evolving uses of space-based PNT services.”
This article identifies and describes several important considerations in the development of GPS spectrum interference standards towards achieving the stated EXCOM goals. These include the identification of characteristics of adjacent band systems and an assessment of the susceptibility of all GPS receiver types towards interference in adjacent bands. Also of vital importance to protecting GPS receivers is an understanding of the user base, applications, and where the receivers for each application may be located while in use. This information, along with the selection of proper propagation models, allows one to establish transmission limits on new adjacent-band systems that will protect currently fielded GPS receivers. The article further comments on the implications of the evolution of GPS and foreign satellite navigation systems upon the development of efficacious spectrum interference standards.
Adjacent Band Characteristics
The type of adjacent-band system for which there is currently the greatest level of interest is a nationwide wireless fourth-generation (4G) terrestrial network to support the rapidly growing throughput demands of personal mobile devices. Such a nationwide network would likely consist of tens of thousands of base stations distributed throughout the United States and millions of mobile devices. The prevalent standard at the present time is Long Term Evolution (LTE), which is being deployed by all of the major U.S. carriers. LTE and Advanced LTE provide an efficient physical layer for mobile wireless services. Worldwide Interoperability for Microwave Access (WiMAX) is a competing wireless communication standard for 4G wireless that is a far-distant second in popularity.
For the purposes of the discussion within this article, an LTE network is assumed with characteristics similar to that proposed by LightSquared but perhaps with base stations and mobile devices that transmit upon different center frequencies and bandwidths. The primary characteristics include:
- Tens of thousands of base stations nationwide, reusing frequencies in a cellular architecture, with the density of base stations peaking in urban areas.
- Base-station antennas at heights from sub-meter to 150 meters above ground level (AGL), with a typical height of 20–30 meters AGL. Each base station site has 1–3 sector antennas mounted on a tower such that peak power is transmitted at a downtilt of 2–6 degrees below the local horizon, with a 60–70 degree horizontal 3-dB beamwidth and 8–9 degree vertical 3-dB beamwidth.
- Peak effective isotropic radiated power (EIRP) in the vicinity of 20–40 dBW (100–10,000 W) per sector.
- Mobile devices transmit at a peak EIRP of around 23 dBm (0.2 W), but substantially lower most of the time when lower power levels suffice to achieve a desired quality of service as determined using real-time power control techniques.
- As LTE uses efficient transmission protocols, emissions can be accurately modeled as brickwall, that is, confined to a finite bandwidth around the carrier.
Throughout this article it will be presumed that LTE emissions in the bands authorized for RNSS systems such as GPS will be kept sufficiently low through regulatory means.
The opening photo shows a typical base-station tower, with three sectors per cellular service provider and with multiple service providers sharing space on the tower, including non-cellular fixed point microwave providers. As a cellular network is being built out, coverage is at first most important, and many base-station sites will use minimum downtilt and peak EIRPs within the ranges described above. As the network matures, capacity becomes more important. High-traffic cells are split through the introduction of more base stations, and this is commonly accompanied by increased downtilts and lower EIRPs.
The assumed characteristics for adjacent band systems plays a paramount role in determining compatibility with GPS, and obviously lower-power adjacent-band systems would be more compatible. If compatibility with GPS precludes 4G network implementation on certain underutilized frequencies adjacent to RNSS bands, then it may be prudent to refocus attention for these bands on alternative lower-power systems.
GPS Receiver Susceptibility
Over the past two years, millions of dollars have been expended to measure or analyze the susceptibility of GPS receivers to adjacent band interference as part of U.S. regulatory proceedings for LightSquared. Measurements were conducted through both radiated (see photo) and conducted tests at multiple facilities, as well as in a live-sky demonstration in Las Vegas. This section summarizes the findings for seven categories of GPS receivers. These categories, which were originally identified in the Federal Communications Commission (FCC)-mandated GPS-LightSquared Technical Working Group (TWG) formed in February 2011, are: aviation, cellular, general location/navigation, high-precision, timing, networks, and space-based receivers.
Aviation. Certified aviation GPS receivers are one of the few receiver types for which interference requirements exist. These requirements take the form of an interference mask (see Figure 1) that is included in both domestic and international standards. Certified aviation GPS receivers must meet all applicable performance requirements in the presence of interference levels up to those indicated in the mask as a function of center frequency. In Figure 1 and throughout this article, all interference levels are referred to the output of the GPS receiver passive-antenna element. Although the mask only spans 1500–1640 MHz, within applicable domestic and international standards the curves are defined to extend over the much wider range of frequencies from 1315 to 2000 MHz.
A handful of aviation GPS receivers were tested against LightSquared emissions in both conducted and radiated campaigns. The results indicated that these receivers are compliant with the mask with potentially some margin. However, the Federal Aviation Administration (FAA) noted the following significant limitations of the testing:
- Not all receiver performance requirements were tested.
- Only a limited number of certified receivers were tested, and even those tested were not tested with every combination of approved equipment (for example, receiver/antenna pairings).
- Tests were not conducted in the environmental conditions that the equipment was certified to tolerate (for example, across the wide range of temperatures that an airborne active antenna experiences, and the extreme vibration profile that is experienced by avionics upon some aircraft).
Due to these limitations, the FAA focused attention upon the standards rather than the test results for LightSquared compatibility analyses, and these standards are also recommended for use in the development of national GPS interference standards. One finding from the measurements of aviation receivers that may be useful, however, is that the devices tested exhibited susceptibilities to out-of-band interference that were nearly constant as a function of interference bandwidth. This fact is useful since the out-of-band interference mask within aviation standards is only defined for continuous-wave (pure tone) interference, whereas LightSquared and other potential adjacent-band systems use signals with bandwidths of 5 MHz or greater.
Cellular. The TWG tested 41 cellular devices supplied by four U.S. carriers (AT&T, Sprint, US Cellular, and Verizon) against LightSquared emissions in the late spring/early summer of 2011. At least one of the 41 devices failed industry standards in the presence of a 5- or 10-MHz LTE signal centered at 1550 MHz at levels as low as –55 dBm, and at least one failed for a 10-MHz LTE signal centered at 1531 MHz at levels as low as –45 dBm. The worst performing cellular devices were either not production models or very old devices, and if the results for these devices are excluded, then the most susceptible device could tolerate a 10-MHz LTE signal centered at 1531 MHz at power levels of up to –30 dBm. Careful retesting took place in the fall of 2011, yielding a lower maximum susceptibility value of –27 dBm under the same conditions.
General Location/Navigation. The TWG effort tested 29 general location/navigation devices. In the presence of a pair of 10-MHz LTE signals centered at 1531 MHz and 1550 MHz, the most susceptible device experienced a 1-dB signal-to-noise ratio (SNR) degradation when each LTE signal was received at –58.9 dBm. In the presence of a single 10-MHz LTE signal centered at 1531 MHz, the most susceptible device experienced a 1-dB SNR degradation when the interfering signal was received at –33 dBm.
Much more extensive testing of the effects of a single LTE signal centered at 1531 MHz on general location/ navigation devices was conducted in the fall of 2011, evaluating 92 devices. The final report on this campaign noted that 69 of the 92 devices experienced a 1-dB SNR decrease or greater when “at an equivalent distance of greater than 100 meters from the LightSquared simulated tower.” Since the tower was modeled as transmitting an EIRP of 62 dBm, the 100-meter separation is equivalent to a received power level of around –14 dBm. The two most susceptible devices experienced 1-dB SNR degradations at received power levels less than –45 dBm.
High Precision, Timing, Networks. The early 2011 TWG campaign tested 44 high-precision and 13 timing receivers. 10 percent of the high-precision (timing) devices experienced a 1-dB or more SNR degradation in the presence of a 10-MHz LTE signal centered at 1550 MHz at a received power level of –81 dBm (–72 dBm). With the 10-MHz LTE signal centered at 1531 MHz, this level increased to –67 dBm (–39 dBm).
The reason that some high-precision GPS receivers are so sensitive to interference in the 1525–1559 MHz band is that they were built with wideband radiofrequency front-ends to intentionally process both GPS and mobile satellite service (MSS) signals. The latter signals provide differential GPS corrections supplied by commercial service providers that lease MSS satellite transponders, from companies including LightSquared.
Space. Two space-based receivers were tested for the TWG study. The first was a current-generation receiver, and the second a next-generation receiver under development. The two receivers experienced 1-dB C/A-code SNR degradation with total interference power levels of –59 dBm and –82 dBm in the presence of two 5-MHz LTE signals centered at 1528.5 MHz and 1552.7 MHz. For a single 10-MHz LTE signal centered at 1531 MHz, the levels corresponding to a 1-dB C/A-code SNR degradation increased to –13 dBm and –63 dBm. The next-generation receiver was more susceptible to adjacent-band interference because it was developed to “be reprogrammed in flight to different frequencies over the full range of GNSS and augmentation signals.”
Discussion. Although extensive amounts of data were produced, the LightSquared studies are insufficient by themselves for the development of GPS interference standards, since they only assessed the susceptibility of GPS receivers to interference at the specific carrier frequencies and with the specific bandwidths proposed by LightSquared. If GPS interference standards are to be developed for additional bands, then much more comprehensive measurements will be necessary.
Interestingly, NTIA in 1998 initiated a GPS receiver interference susceptibility study, funded by the Department of Defense (DoD) and conducted by DoD’s Joint Spectrum Center. One set of curves produced by the study is shown in Figure 2. This format would be a useful output of a further measurement campaign. The curves depict the interference levels needed to produce a 1-dB SNR degradation to one GPS device as the bandwidth and center frequency of the interference is varied. The NTIA curves only extended from GPS L1 (1575.42 MHz) ± 20 MHz. A much wider range would be needed to develop GPS interference standards as envisioned by the PNT EXCOM. It may be possible, to minimize testing, to exclude certain ranges of frequencies corresponding to bands that stakeholders agree are unlikely to be repurposed for new (for example, mobile broadband) systems.
The LightSquared studies, with the exception of those focused on aviation and space applications, spent far less attention to receiver-transmitter proximity. Minimum separation distances and the associated geometry are obviously very important towards determining the maximum interference level that might be expected for a given LTE network (or other adjacent band system) laydown.
Within the TWG, the assumption generally made for other (non-aviation, non-space) GPS receiver categories was that they could see power levels that were measured in Las Vegas a couple of meters above the ground from a live LightSquared tower. Figure 3 shows one set of received power measurements from Las Vegas. In the figure, the dots are measured received power levels made by a test van. The top curve is a prediction of received power based upon the free-space path-loss model. The bottom curve is a prediction based upon the Walfisch-Ikegami line-of-sight (WILOS) propagation model. The NPEF studies presumed that the user could be within the boresight of a sector antenna even within small distances of the antenna (where the user would need to be at a significant height above ground).
The difference between the above received LTE signal power assumptions has been hotly debated, especially after LightSquared proposed limiting received power levels from the aggregate of all transmitting base stations as measured a couple of meters above the ground in areas accessible to a test vehicle. After summarizing the aviation scenarios developed by the FAA, this section highlights scenarios where so-called terrestrial GPS receivers can be at above-ground heights well over 2 meters. The importance of accurately understanding transmitter-receiver proximity is illustrated by Figure 4. This shows predicted received power levels for one LTE base station sector transmitting with an EIRP of 30 dBW and with an antenna height of 20 meters (65.6 feet). The figure was produced assuming the free-space path-loss model and a typical GPS patch-antenna gain pattern for the user. Note that maximum received power levels are very sensitive to the victim GPS receiver antenna height.
Aviation. The first LightSquared-GPS study conducted for civil aviation was completed by the Radio Technical Commission for Aeronautic (RTCA) upon a request from the FAA. Due to the extremely short requested turnaround time (3 months), RTCA consciously decided not to devote any of the available time developing operational scenarios, but rather re-used scenarios that it had developed for earlier interference studies. It was later realized that the combination of five re-used scenarios and assumed LightSquared network characteristics did not result in an accurate identification of the most stressing real-world scenarios. For instance, within the RTCA report, base stations’ towers were all assumed to be 30 meters in height. At this height, towers could not be close to runway thresholds where aircraft are flying very low to the ground, because this situation would be precluded by obstacle clearance surfaces. Later studies used actual base-station locations, from which the aviation community became aware that cellular service providers do place base stations close to airports by utilizing lower base-station heights as necessary to keep the antenna structure just below obstacle clearance surfaces.
The FAA completed an assessment of LightSquared-GPS compatibility in January 2012 that identified scenarios where certified aviation receivers could experience much higher levels of interference than was assessed in the RTCA report. The areas where fixed-wing and rotary-wing aircraft rely on GPS are depicted in Figures 5 and 6 (above the connected line segments), respectively.
Aircraft rely upon GPS for navigation and Terrain Awareness and Warning Systems (TAWS). Helicopter low-level en-route navigation and TAWS for fixed- and rotary-wing aircraft are perhaps the most challenging scenarios for ensuring GPS compatibility with adjacent-band cellular networks. In these scenarios, the aircraft can be within the boresight of cellular sector antennas and in very close proximity, resulting in very high received-power levels. The FAA attempted to provide some leeway for LightSquared while maintaining safe functionality of TAWS through the concept of exclusion zones (see Figure 7). The idea of an exclusion zone is that, at least for cellular base-station transmitters on towers that are included within TAWS databases, that it would be permitted for the GPS function to not be available for very small zones around the LTE base-station tower. This concept is currently notional only; the FAA plans to more carefully evaluate the feasibility of this concept and appropriate exclusion-zone size with the assistance of other aviation industry stakeholders.
High-precision and Networks: Reference Stations. To gain insight into typical reference-station heights for differential GPS networks, the AGL heights of sites comprising the Continuously Operating Reference Station (CORS) network organized by the National Geodetic Survey (NGS) were determined. The assessment procedure is detailed in the Appendix.
Figure 8 portrays a histogram of estimated AGL heights for the 1543 operational sites within the continental United States (CONUS) as of February 2012. The accuracy of the estimated AGL heights is on the order of 16 meters, 90 percent, limited primarily by the quality of the terrain data that was utilized. The mean and median site heights are 5.7 and 5.2 meters, respectively.
RALR, atop the Archdale Building in Raleigh, North Carolina, was the tallest identified site at 64.1 meters. This site, however, was decommissioned in January 2012 (although it was identified as operational in a February 2012 NGS listing of sites). The second tallest site identified is WVHU in Huntington, West Virginia at 39.6 meters, which is still operational atop of a Marshall University building. 223 of the 1543 CORS sites within CONUS have AGL heights greater than 10 meters, and furthermore the taller sites tend to be in urban areas where cellular networks tend to have the greatest base-station density.
High Precision and Networks: End Users. Many high-precision end users employ GPS receivers at considerable heights above ground. For instance, high-precision receivers are relied upon within modern construction methods. The adjacent photos show GPS receivers used for the construction of a 58-story skyscraper called The Bow in Calgary, Canada. For this project, a rooftop control network was established on top of neighboring buildings using both GPS receivers and other surveying equipment (for example, 360-degree prisms for total stations), and GPS receivers were moved up with each successive stage of the building to keep structural components plumb and properly aligned. Similar techniques are being used for the Freedom Tower, the new World Trade Center, in New York City, and many other current construction projects.
Other terrestrial applications that rely on high-precision GPS receivers at high altitudes include structural monitoring and control of mechanical equipment such as gantry cranes. At times, even ground-based survey receivers can be substantially elevated. Although a conventional surveying pole or tripod typically places the GPS antenna 1.5 – 2 meters above the ground, much longer poles are available and occasionally used in areas where obstructions are present. 4-meter GPS poles are often utilized, and poles of up to 40 ft (12.2 meters) are available from survey supply companies.
General Location/Navigation. Although controlling received power from a cellular network at 2 meters AGL may be suitable to protect many general navigation/location users, it is not adequate by itself. For example, GPS receivers are used for tracking trucks and for positive train control (the latter mandated in the United States per the Rail Safety Improvement Act of 2008). GPS antennas for trucks and trains are often situated on top of these vehicles. Large trucks in the United States for use on public roads can be up to 13 ft, 6 in (~4.1 meters), and a typical U.S. locomotive height is 15 ft, 5 in (~4.7 meters). Especially in a mature network that is using high downtilts, received power at these AGL heights can be substantially higher than at 2 meters.
Within the TWG and NPEF studies, the general location/navigation GPS receiver category is defined to include non-certified aviation receivers. One notable application is the use of GPS to navigate unmanned aerial vehicles. UAVs are increasingly being used for law enforcement, border control, and many other applications where the UAV can be expected to occasionally pass within the boresight of cellular antennas at short ranges.
Cellular. The majority of Americans own cell phones, and a growing number are using cell phones as a replacement for landlines within their home. Already, 70 percent of 911 calls are made on mobile phones. Although pedestrians and car passengers are often within 2 meters of the ground, this is not always the case. Figure 9 shows three cellular sector antennas situated atop a building filled with residential condominiums. The rooftop is accessible and frequently used by the building inhabitants. According to an online real estate advertisement, “The Garden Roof was voted the Best Green Roof in Town and provides amazing 360 degree views of downtown Nashville as well as four separate sitting areas and fabulous landscaping.” One of the sector antennas is pointing towards the opposite corner of the building. If the downtilt is in the vicinity of 2–6 degrees, then it is quite likely that a person making a 911 call from the rooftop could see a received power level of –10 dBm to 0 dBm, high enough to disrupt GPS within most cellular devices if the antennas were transmitting in the 1525–1559 MHz band.
This situation is not unusual. Many cellular base stations are situated on rooftops in urban areas, and many illuminate living areas in adjacent buildings. In recent years, New York City even considered legislation to protect citizens from potential harmful effects of the more than 2,600 cell sites in the city, since many sites are in very close proximity to residential areas.
Within the LightSquared proceedings, there was a tremendous amount of debate regarding propagation models. Communication-system service providers typically use propagation models that are conservative in their estimates of received power levels in the sense that they overestimate propagation losses. This conservatism is necessary so that the service can be provided to end users with high availability. From the standpoint of potential victims of interference, however, it is seen as far more desirable to underestimate propagation losses so that interference can be kept below an acceptable level a very high percentage of time. As shown in Figure 3, some received power measurements from the Las Vegas live-sky test indicate values even greater than would be predicted using free-space propagation model. Statistical models that allow for this possible were used in the FAA Status Report. The general topic of propagation models is worthy of future additional study if GPS interference standards are to be developed.
GPS is being modernized. Additionally, satellite navigation users now enjoy the fact that the Russian GLONASS system has recently returned to full strength with the repopulation of its constellation. In the next decade, satellite navigation users also eagerly anticipate the completion of two other global GNSS constellations: Europe’s Galileo and China’s Compass. Notably, between the GPS modernization program and the deployment of these other systems, satellite navigation users are expected to soon be relying upon equipment that is multi-frequency and that needs to process many more signals with varied characteristics. New equipment offers an opportunity to insert new technologies such as improved filtering, but of course the need to process additional signals and carrier frequencies may make GNSS equipment more susceptible to interference as well. Clearly, these developments will need to be carefully assessed to support the establishment of GPS spectrum interference standards.
This article has identified a number of considerations for the development of GPS interference standards, which have been proposed by the PNT EXCOM. If the United States proceeds with the development of such standards, it is hoped that the information within this article will prove useful to those involved.
Appendix: AGL Heights of CORS Network Sites
The National Geodetic Survey Continuously Operating Reference Station (CORS) website provides lists of CORS site locations in a number of different reference frames. To determine the height above ground level () for each site within this study, two of these files (igs08_xyz_comp.txt and igs08_xyz_htdp.txt) were used. These two files provide the (x,y,z) coordinates of the antenna reference point (ARP) for each site in the International GNSS Service 2008 (IGS08) reference frame, which is consistent with the International Terrestrial Reference Frame (ITRF) of 2008. These coordinates are divided into two files by NGS, since the site listings also provide site velocities and velocities are either computed (for sites that have produced data for at least 2.5 years) or estimated (for newer sites). The comp file includes sites with computed velocities and the htdp file includes sites with estimated velocities (using a NGS program known as HTDP).
The data files can be used to readily produce height above the ellipsoid, , for each site. This height can be found using well-known equations to convert from (x, y, z) to (latitude, longitude, height). Obtaining estimates of requires information on the geoid height and terrain data, per the relationship:
For the results presented in this article, terrain data was obtained from http://earthexplorer.usgs.gov in the Shuttle Radar Topography Mission (SRTM) Digital Terrain Elevation Data (DTED) Level 2 format. For this terrain data, the horizontal datum is the World Geodetic System (WGS 84). The vertical datum is Mean Sea Level (MSL) as determined by the Earth Gravitational Model (EGM) 1996. Each data file covers a 1º by 1º degree cell in latitude/longitude, and individual points are spaced 1 arcsec in both latitude and longitude. The SRTM DTED Level 2 has a system design 16 meter absolute vertical height accuracy, 10 meters relative vertical height accuracy, and 20 meter absolute horizontal circular accuracy. All accuracies are at the 90 percent level. Considering the accuracies of the DTED data, the differences between WGS-84 and IGS08 as well as between the ARP and antenna phase center were considered negligible. Geoid heights were interpolated from 15-arcmin data available in the MATLAB Mapping Toolbox using the egm96geoid function.
Lower AGL heights are preferred for CORS sites to minimize motion between the antenna and the Earth’s crust. However, many sites are at significant heights above the ground by necessity, particularly in urban areas due to the competing desire for good sky visibility.
Christopher J. Hegarty is the director for communications, navigation, and surveillance engineering and spectrum with The MITRE Corporation. He received a D.Sc. degree in electrical engineering from George Washington University. He is currently the chair of the Program Management Committee of the RTCA, Inc., and co-chairs RTCA Special Committee 159 (GNSS). He is the co-editor/co-author of the textbook Understanding GPS: Principles and Applications, 2nd Edition.