Butler National Corporation, which specializes in the aerospace sector of structural modification, maintenance, repair, and overhaul, announces issuance of the Supplemental Type Certificate (STC) for installation of the new Garmin GTN series of navigators that provide GPS, navigation, and communications. The installation is for GTN 750 navigators in the Learjet Models 35/35A/36/36A with the FC-200 autopilot and the Learjet Model 24.
The Garmin GTN series features intuitive touchscreen controls and a large-screen display that give Learjet pilots unprecedented access to high-resolution mapping, graphical flight planning, and geo-referenced charting, among many other features. The installation also features new GPS roll-steering that allows seamless navigation operations with turn anticipation and waypoint sequencing interfaced to the autopilot.
“This approval offers a significant and economical avionics upgrade for the Lear 30 series airplanes,” said Clark Stewart, president and CEO of Butler. “The STC for the new Garmin GTN series allows us to tap into a sizeable upgrade market for retrofit of flight management systems. The Garmin GTN upgrade provides significant functionality upgrades, including WAAS GPS approaches and roll-steering interface to the autopilot.”
“We have designed the installation to provide cost-effective options to meet the various Learjet operator requirements. We will be offering the GTN Learjet installations through our avionics facility Kings Avionics starting under $100,000,” commented Craig Stewart, Aerospace division president.
Butler National Corporation operates in the Aerospace and Services business segments. The Aerospace segment focuses on the manufacturing of support systems for commercial and military aircraft including the Butler National TSD for the Boeing 737 and 747 Classic aircraft, switching equipment for Boeing McDonnell Douglas Aircraft, weapon control systems for Boeing Helicopter, and performance enhancement structural modifications for Learjet, Cessna, Dassault, and Beechcraft business aircraft.
The Federal Aviation Administration (FAA) has awarded Raytheon Company a two-year contract extension to continue to provide services for the Wide Area Augmentation System (WAAS), a safety system that provides satellite-based navigation in the continental United States, Alaska, Canada and Mexico. The $30.1 million contract extends the period of performance through Sept. 24, 2013.
“Raytheon has been the FAA’s prime contractor for WAAS since the system was commissioned for operational use in the United States in 2003,” said Michael Prout, vice president of Security and Transportation Systems for Raytheon’s Network Centric Systems business. “The contract marks another milestone in the continuing partnership between Raytheon and the FAA to improve safety and efficiency for pilots.”
According to the announcement, Raytheon will provide life-cycle support and other services to improve service reliability and availability, and increase the coverage area through system enhancements. WAAS enables GPS to meet air navigation performance and safety requirements for en route, terminal, non-precision approach, and approach with vertical guidance operations.
When jamming interfered with GPS signals at Newark Airport, a three-month effort determined that low-power, mobile personal-privacy devices were responsible. This article describes how they were found and outlines how the observable parameters of such devices encompass a wide variation in RF spectra and internal modulation.
Personal privacy devices (PPDs) are now recognized as being responsible for causing interference to GPS receivers. However, in November of 2009, when the Local-Area Augmentation System (LAAS) installed its first Ground-Based Augmentation System (GBAS) at Newark International Airport (EWR), this fact was not known. Within the first month of its installation, anomalies in GBAS processing were correlated to the presence of radio-frequency interference (RFI). Initial efforts to determine the source of this unexpected RFI were not successful.
The Federal Aviation Administration (FAA) had a significant interest in finding this RFI source, leading to deployment of RFI detection and location equipment by several groups. Zeta Associates temporarily installed equipment in early January 2010 that was capable of detecting and characterizing RFI but did not have emitter location capability.
Determining that an RFI transmitter is in motion is more certain if the RFI is observed simultaneously by multiple sensors. However, analysis from hundreds of RFI events indicates that when an RFI source is in motion, observations collected by a single sensor can provide sufficient information to determine that the RFI transmitter is in motion.
Continuing interest in understanding PPD effects on GPS receivers led to the installation of remotely accessible monitoring equipment that provides detailed characteristics of these devices. Remote access facilitates monitoring, particularly since PPDs are present for 30 to 60 seconds at a time and only a few times a day.
Background
The January 2010 deployment included a WAAS GPS receiver, spectrum analyzer, and a Zeta custom-developed Snapshot System, assembled from commercial off-the-shelf (COTS) equipment for conducting WAAS site-installation surveys, and capable of capturing intermittent short-duration RFI events. It consists of a tuneable receiver (10 MHz to 3 GHz) whose RF front end spans 25 MHz that is digitized at a sample rate of 56 MHz, with storage capacity sufficient for up to 80 minutes. Once captured, the time-series data can be analyzed in many different ways. Possible analysis techniques include examination of the raw time samples, generation of spectral plots, or demodulation of the RFI signal. Each approach can lead to a better understanding of the underlying interference signal. If digital data is present and can be demodulated, it might be possible to associate the demodulated bits with a known transmitter.
Data can be captured manually or programmatically using a trigger determined by an algorithm that monitors WAAS GPS receiver automatic gain control (AGC) logs. The AGC function within a WAAS receiver has a well-behaved response for normal Gaussian noise RF environments. When RFI is present, the AGC exhibits atypical responses that then trigger the Snapshot System. As WAAS receivers utilize both L1 and L2, and each RF path has its own AGC, it is possible to detect the presence of potential RFI at either L1 or L2.
The Snapshot System RF input for this deployment was from a PCTEL antenna identical to those used at WAAS reference sites. This antenna incorporates a triplexer that provides three separate 40 MHz passbands each centered on L1, L2, and L5, with approximately 50 dB of gain. This antenna was located approximately one mile north of the four LAAS antennas within Port Authority of New York and New Jersey (PANYNJ) Building 80.
Within the first hour of being deployed on January 20, 2010 the Snapshot System had detected and captured one RFI event in the GPS L1 band. After one day, the Snapshot System had detected and captured more than 25 separate instances of RFI within the GPS L1 band. Most RFI events were narrowband (10s of kHz bandwidth) and short duration (no more than 3 seconds).
However, there also were five RFI events that spanned more than 15 MHz across L1 (Figure 1) were present as long as 20 seconds and at a power level as much as 25 dB above the receive antenna’s noise floor. Some of these RFI events were strong enough to reduce a WAAS G-II receiver C/N0 by as much as 20 dB and thereby resulted in loss of tracking for lower-elevation GPS satellites. Higher-elevation GPS satellites were able to continue tracking throughout these events but at a lower C/N0. The wideband RFI events were also detected by the SLS 4000 GBAS monitor and coincided with tracking problems in the LAAS GBAS receivers.
FIGURE 1. EWR wideband RFI.
Two of the captured broadband RFI events were demodulated and analyzed. The underlying linear frequency modulation (FM) signal swept over more than 15 MHz in less than 1 millisecond (Figure 2).
FIGURE 2. FM demodulated wideband RFI.
At that time, it was not known if the source of the RFI was stationary or moving, whether it was unintentional (emanating from a licensed transmitter but with malfunctioning electronics), inadvertent (equipment normally used for test purposes and capable of operating in the GPS band but accidentally left on), or intentional (purposeful jamming of GPS).
Since the RFI was observed by GPS receivers separated by 1,700 meters, a search was undertaken to identify any other GPS receivers in the vicinity of EWR. One National Geodetic Survey (NGS) continuously operating reference site (CORS) NJI2 is located near EWR about 4,500 meters northwest from Building 80. Analysis of data from NJI2 during the same time periods that RFI was detected by the WAAS and LAAS receivers did not contain any indication of RFI, and therefore suggested that the source of RFI was more localized to EWR.
The Snapshot System remained in place for approximately two weeks before moving to another location. Collected data was analyzed, showing that wideband RFI was associated with significant degradation to both the WAAS and LAAS receivers. Additional characteristics noted the RFI was intermittent, lasting typically 30 seconds but no more than 60 seconds, was observed more often Monday through Friday, and most frequently around 8 a.m. local time.
Locating The RFI
Figure 3 shows a Google map of EWR with blue dots indicating the location of the four LAAS antennas, a green dot for Building 80, and a yellow dot for the GBAS shelter. EWR is adjacent to the New Jersey Turnpike (NJT), which has seven southbound and seven northbound lanes of traffic.
FIGURE 3 Google map of EWR.
Since the Snaphsot System did not include location capability, other teams with direction-finding equipment, including beam-forming antennas, travelled to EWR to try to locate the RFI source. These teams were on site at various times from February to March. However, those efforts did not provide sufficiently reliable information to reduce the search area. By mid-March, the search area remained identical to that of January.
Zeta then deployed two WAAS G-II receivers separated by considerable distance (1,722 meters) to monitor for RFI, and analyze each receiver’s response only when RFI sufficient to significantly degrade GPS reception was detected. One receiver was located within Building 80, and the second receiver within the GBAS shelter near the LAAS antennas. This configuration was designed to determine degradation relative to each reference receiver and thereby establish probable search areas for the RFI emitter. The Zeta equipment also incorporated a rotating directional antenna (at the GBAS shelter shown in Figure 4) that was commanded to rotate only when significant RFI was detected.
FIGURE 4A. Antennas on roof of GBAS shelter.FIGURE 4B. Antennas on roof of GBAS shelter.
The expectation was that RFI would be detected simultaneously by both GPS receivers, and that the relative degradation in normalized C/ N0 would provide an indication as to which location lay in closer proximity to the RFI source. The rotating high-gain directional antenna would then indicate a reduced probable search area consistent with the relative degradation between the two receivers. At the time this equipment was deployed, it was still thought that the RFI was most likely stationary and high-power. However, the measurement results were quite different than expected. Subsequent data analysis from this equipment revealed that the RFI was low-power and moving, specifically moving along the NJT.
The Zeta equipment was deployed on March 19, 2010, and remained in place while operating automatically. On March 25, data collected during the previous week were analyzed. During this 1-week collection there were 11 instances when both receivers detected wideband RFI events and one antenna rotation even partially tracked one wideband RFI emitter. Such data was indicative of a non-stationary emitter, a finding that was quite significant. Based on data from the two receivers, the apparent velocity of the RFI emitters ranged between 45 miles per hour (mph) to 72 mph. Initial analysis of antenna-rotation data also indicated the RFI source was east of the GBAS shelter and moving south on the NJT.
Understanding the importance of degradations from both receivers was crucial in determining that the RFI has attributes of transmitting at low power and is moving. Had a single stationary RFI emitter been responsible for these observations, the degradations measured at each receiver would have occurred at essentially the same time, not 50 to 80 seconds apart. A high-power moving RFI emitter would also have produced degradations at both receivers at the same time, and since that was not observed, the conclusion was that the RFI emitter was relatively low in output power. Low-power RFI emitters will cause significant degradation to GPS receivers only when they are in close proximity to them, on the order of hundreds of meters.
Receiver data logs were processed specifically for degradation in normalized C/N0. Normalized C/N0 was only computed for those satellites above 20 degrees, and all of those results were averaged together. Prior knowledge regarding WAAS PCTEL antennas has established an expected C/N0 versus satellite elevation that is accurate to approximately ±1 dB with a nominal mean of 0 dB. This normalized C/N0 represents an average of all satellites in view. However, individual satellite signal strength can vary greater than ± 1 dB. Significant deviations of more than –3 dB are indicative of strong RFI within the GPS processing band. Normalized C/N0 was plotted for each day that data was collected, followed by expanding those time periods where significant degradation was present.
Figure 5 shows data of the first evidence of a low-power moving PPD. Data for Building 80 receiver is in blue and data from the GBAS shelter receiver in pink. Since Building 80 is north of the GBAS shelter, when degradations occurred first at Building 80, this implies that the RFI emitter is moving from north to south. Similarly, when degradations were first seen at the GBAS shelter, the RFI emitter was moving from south to north. This plot uses major time grids of 60 seconds and minor grids of 10 seconds.
FIGURE 5. Normalized C/N0 observed at Building 80 and GBAS shelter.
The double separate degradations observed by the Building 80 receiver have only been observed from monitoring equipment located at that building, and have since been associated with travel paths of PPDs on the nearby highways. Both GPS receiver and spectral data contain this same characteristic. This characteristic is due to the fact that vehicles traveling south on the NJT have clear line of sight to the roof of Building 80 (shown by Figure 6) before they travel under Interstate 78, after which they pass next to Building 80. During the time that they are under Interstate 78, their transmissions are blocked in the direction of the roof of Building 80.
FIGURE 6A. View of NJT near Building 80.FIGURE 6B. View of NJT near Building 80.
Spectral data as observed by the 4-foot reflector is shown in Figure 7. Figure 8 shows spectral maximum data as collected by the 4-foot linearly polarized reflector along with additional information.
FIGURE 7. Wideband RFI observed by 4-foot reflector (click to enlarge.)FIGURE 8A. Pink represents spectral maximum data as observed through the reflector, green represents the azimuth of that antenna, and blue the reported degradation of the GBAS shelter receiver.FIGURE 8B. Pink represents spectral maximum data as observed through the reflector, green represents the azimuth of that antenna, and blue the reported degradation of the GBAS shelter receiver.
When the GBAS shelter receiver at first detected RFI, the reflector began rotating from an azimuth of 0 degrees in a clockwise direction. At the same time, a spectrum analyzer began capturing spectra at a rate of 3 per second. The first spectral maximum was observed at an azimuth of 30 degrees, a direction in which the antenna was pointed towards the NJT, to a location approximately 900 meters away from the GBAS shelter. The next time spectra were at high levels occurred for azimuths between 145 to 195 degrees, or southeast of the GBAS shelter. The approach of using a rotating antenna was originally intended to provide a direction towards a stationary source and not to track a moving emitter. However, it appears that to some extent, the rotating antenna in fact did track a moving emitter from north to south.
On the afternoon of the day these results were communicated to the FAA lead for the EWR RFI investigation, all search activities were shifted to the NJT and away from the airport operating area. Just south of the GBAS shelter there is an official-use overpass that straddles the NJT. All detection equipment was positioned onto the overpass, under the hypothesis that the RFI was emanating from vehicles traveling the NJT. Evidence substantiating this initial finding was found within a day, and approximately one month later a concerted effort was undertaken to identify and stop a single vehicle that was using a PPD.
The Zeta equipment remained in place for many months and continued to provide additional evidence of PPD characteristics. Early in the investigation it was hoped that only a few PPDs had been responsible, but as more data was collected it became evident that many different types of PPDs were traveling along the NJT past EWR.
Modeling PPD Effect
Once it was realized that the RFI was from low power moving emitters, a simple model was used to predict their degradation effect on WAAS GPS receivers. The model shown in Figure 9 was used for the purpose of computing distance between the RFI emitter and a WAAS antenna and to then compute the additional level of interference noise power that the WAAS antenna would receive. Here, the WAAS PCTEL antenna is located 50 meters from a road that is 2000 meters long and straight and has an RFI emitter transmitting +25 dBm, moving at 32.5 meters per second (72.5 mph) and with clear line of sight to the WAAS antenna.
FIGURE 9. Simple model of moving emitter.FIGURE 10. Model of normalized C/No due to a PPD.
With these assumptions it is a simple matter to compute the additional noise power at the WAAS antenna. Non-coherent summation of the RFI noise and inherent system noise was used to compute the total noise power and therefore the additional degradation in C/N0. The resulting predicted degradation was overlaid on one of the actual RFI events and is shown as a green line in Figure 10. The predicted degradation closely resembles actual event data logged by receivers.
The shape of degradation in normalized C/N0 versus time has been observed in nearly all of the EWR RFI events that have been analyzed. The magnitude of degradation depends on the power of the RFI and its proximity to the GPS antenna, while its time duration depends on the velocity of the vehicle carrying the PPD. The shape is directly related to the distance versus time between the vehicle and the WAAS antenna. Faster/slower moving vehicles with PPDs will simply shrink/stretch the time scale. Curved roadways would have different shapes that could also be readily predicted.
CORS data was revisited after realizing that PPDs were traveling the NJT. Specifically, two CORS sites CTDA (70 meters from Interstate 95) and NJDY (380 meters from Interstate 95) were identified. Data from those two sites were analyzed for a couple of weekdays. Possible evidence of PPDs was found within that data. Reported Signal to Noise Ratio (SNR) from CTDA and NJDY contained variations similar to those observed by GPS receivers at EWR during times when PPD induced RFI has been detected.
Continued RFI Monitoring
The LAAS program desired continued monitoring of RFI from PPDs near EWR, including estimates of their effective isotropic radiated power (EIRP). Additional equipment was assembled to provide this capability and installed on March 3, 2011. This monitoring equipment is located within the GBAS shelter at EWR and comprises several COTS components that incorporate improvements beyond the first Snapshot System used at EWR. Improvements include an upgraded Snapshot System (Figure 11), an RHCP directional antenna, and a wireless modem that provides remote access to the monitoring equipment.
Figure 11A. Snapshot System ICEPOD6-M5.FIGURE 11B. Snapshot System laptop.
Remote access makes it possible to analyze captured RFI data from any computer connected to the Internet, and to modify software if necessary. The new equipment configuration, specifically the use of an AEL AST 1507AA RHCP antenna, was chosen with the explicit purpose of establishing more accuracy in estimated EIRP.
Analysis of data collected during 2010 indicated three significant sources of error in estimating EIRP; Free Space Loss (FSL) from not knowing the exact position of the PPD on the NJT, polarization mismatch loss between the PPD antenna and the receive antenna, and the effects due to transmission from within a vehicle. Differences in FSL loss between the closest southbound lane on the NJT and the most distant northbound lane is on the order of 11 dB. If it is known that the PPD is traveling south, the difference in FSL between the nearest to farthest southbound lanes is less but still about 6 dB. FSL differences for northbound lanes are smaller, on the order of 3 dB. Knowing the direction of travel reduces this uncertainty but does not eliminate it. The WAAS PCTEL antenna is RHCP, but at the horizon has an axial ratio of approximately 5 dB. Many PPDs appear to be using quarter wavelength dipole antenna that are mounted with a rotatable connector. Assuming all PPD antennas are linearly polarized, there would then be an uncertainty of about 5 dB when using data observed by the WAAS PCTEL antenna. The AEL antenna has an axial ratio of 0.2 dB at L1 and significantly reduces uncertainty of polarization mismatch loss. Even though the AEL antenna has a 3 dB mismatch loss, that loss has an uncertainty of 0.2 dB for any orientation of the PPD antenna. The PCTEL antenna has a symmetric gain response relative to the NJT. The AEL antenna was pointed to the north intentionally so that observed RFI power will be stronger when a PPD is north of the GBAS shelter. Simultaneous capture of real time samples from both antennas by the ICEPOD6-M5 provides an indication of the direction in which the PPD is traveling. The third uncertainty comes from the effects of the PPD being located within a vehicle. Vehicular effects on the PPD transmitters were not accounted for due to difficulty in using a simple model. Aloi [1] has measured vehicular effects of quarter wavelength dipole antenna (typically used by PPDs) and has previously published [1] [2] on the effect of vehicles on GPS signals. His most recent findings have not yet been published but indicate that the type of vehicle, the location of a dipole antenna within it, and the position outside of the vehicle from which power is being measured, can lead to significant variation (10 to 15 dB) of the observable power. Consequently, the reported EIRP estimates from the Snapshot System have been referenced to a point just outside the vehicle and do not attempt to account for vehicular effects.
Figure 12 shows a block diagram of the current monitoring equipment. The PCTEL antenna along with the WAAS G-II receiver is used for monitoring GPS signals. WAAS G-II receiver logs are monitored continuously and saved to disk, but snapshots are only taken if the RFI algorithms indicate that RFI is present. A snapshot will be taken if any of three tests (based on AGC, normalized C/N0, or spectral power) are exceeded. Narrowband RFI tends to most impact the AGC response, while wideband RFI tends to trigger the normalized C/N0 metric. Since the Snapshot System is also continuously monitoring the RF spectra from both antennas, an additional test checks if there are significant spectra changes.
FIGURE 12. Current Snapshot monitoring block diagram.
The AEL antenna is connected to a filter/low noise amplifier (LNA) (Delta Microwave L5995) identical to those within the PCTEL antenna, before it is connected to one of the two L-band tuners (900 to 2200 MHz) of the ICE-Online ICEPOD6-M5, which includes 1 TB of long-term storage and 8 GB of high-speed RAM and is capable of sustained data transfers of as high as 400 MB/s between the L-band tuners and disk. Sample rates and RF filtering are programmable and have been set to use a complex sample rate of 40 MSPS and an RF bandwidth of 30 MHz. When RFI is present, data transfers are 320 MB/s. The internal 1 TB disk can store approximately 100 minutes of RFI. Software parameters limit RFI data capture for any single event to no more than 90 seconds. Implementation of a circular buffer within the high-speed RAM (8 seconds for each L-band tuner path) allows continuous capture of RF data while waiting for a trigger indicating that RFI is present. To reduce false alarms, RFI must be present for at least 4 seconds before data is captured. However, no RFI data is lost, because the circular buffer is longer than 4 seconds. RFI data captures typically contain 3 seconds of data at the beginning, with no RFI, and therefore make it possible to observe the onset of the RFI.
The new equipment has captured hundreds of RFI events, spanning a wide range in bandwidth (7 MHz to an estimated 150 MHz), chirp rate (9 kHz to 170 kHz), and power levels (–10 dBm to as much as +20 dBm). Accurately estimating EIRP of moving emitters is a challenge and requires detailed knowledge of the characteristics of all components used in generating the estimate. Furthermore, distance to the RFI source can only be inferred, since its movement precludes exact measurement, and consequently there will always be some uncertainty in any reported EIRP. However, even with these qualifications, there is evidence from Snapshot System data that some of the RFI sources are transmitting at power levels as much as +20 dBm.
Modeling Antenna Responses
The use and orientation of the two antennas was chosen to determine the direction that PPDs are traveling and thereby reduce some of the uncertainty with respect to their exact location. Figure 13 shows the direction (58 degrees) in which the AEL antenna is pointed.
FIGURE 13. Pointing direction of the AEL antenna.
The combination of its beam width (70 degrees) and axial ratio (0.2 dB) results in a nearly uniform gain across all lanes north of the GBAS shelter. Although FSL still depends on the exact location of the PPD, this approach does reduce many of the uncertainties associated with estimating EIRP. Since the PCTEL antenna has an omnidirectional pattern it will have a symmetrical response.
To determine the time that maximum PPD power would be observed by each antenna, a model was used that assumed the PPD was transmitting at constant power, with fixed polarization, travelling at a constant velocity, and there were no obstacles between the transmitter and each antenna. FSL was calculated for each second of travel and used to determine the magnitude of RFI power that would be received. This value was then used to calculate a nominal Interference to Signal (I/S) relative to the GPS signal. The distance between each antenna and each of the travel lanes on the NJT were also used. The intent of this model was to understand how I/S would vary with time for each of the NJT travel lanes. Figure 14 shows the predicted I/S for the PCTEL antenna and Figure 15 for the AEL antenna. In each of these plots red, yellow and orange represent 3 of the 7 south bound travel lanes and green, blue and purple represent 3 of the 7 north bound travel lanes. A time of 0 was used for the time when the PPD is nearest physically to the GBAS shelter. A nominal velocity of 30 meters/second (67 MPH) was used for the PPD and I/S was computed for 30 seconds before and after its closest approach (± 900 meters north and south of the GBAS shelter). If the PPD travels slower than 30 m/s then the following curves would be wider for the same times. Similarly, if the PPD travels faster, these same curves would be narrower.
FIGURE 14. Predicted I/S for PCTEL antenna (click to enlarge.)FIGURE 15. Predicted I/S for AEL AST-1507AA (click to enlarge.)
Modeling of the AEL antenna took into account its pattern and orientation. It has less gain towards the south, and consequently observed power from a PPD located south of the GBAS shelter is much less. A southbound PPD will initially be within the main beam of the AEL antenna; therefore the expected interference to signal ratio (I/S) will gradually increase until it passes to the south of the GBAS shelter. Similarly, a northbound PPD will not exhibit significant I/S until north of the GBAS shelter. Figure 15 indicates that the maximum I/S occurs within 2 seconds of the point where the PPD is closest to the GBAS shelter. Typical GPS receivers can tolerate an I/S of 30 dB for CW type signals.
Processed RFI data does display some of these characteristics but with some important differences. PPD power was measured once every millisecond using Snapshot System data and total power within the bandwidth of that PPD was calculated. Total power from both the PCTEL (green) and AEL (pink) antennas were then plotted together with one example for a southbound PPD in Figure 16 and a northbound PPD in Figure 17.
FIGURE 16. RFI Power of southbound PPD (click to enlarge.)FIGURE 17. RFI Power of northbound PPD (click to enlarge.)
Although the envelope of the measured average power tends to have the shape that modeling predicts, there are significant variations over short periods of time. Figure 18 expands a portion of one example and indicates that RFI power varied by more than 17 dB in 0.2 seconds. Examination of spectral data for time intervals of less than one second frequently contains significant changes in observable power. Swept CW from PPDs should exhibit relatively flat RF spectral power, but typical observed spectra include sloping across the band and notches. Possible explanations for these observations include: blockage and diffraction from other vehicles near the one containing the PPD, multipath from other vehicles on the NJT, and the effect of transmitting from within a vehicle. Although some of the snapshot captures exhibit smooth power variation similar to predicted, the vast majority of the hundreds of snapshots exhibit significant variations in power.
FIGURE 18. RFI Power of southbound PPD (expanded).
Examples of Observed PPDs
The variety of PPDs observed by the updated EWR monitoring equipment has been surprising. Within its first month of operation, more than 40 PPDs were observed with no less than 19 from unique and different PPD transmitters. Classification of PPD transmitters is based on the combination of RF spectra and the spectrum of the FM demodulated data. Although the observed PPD transmitters use a linearly swept FM sawtooth, most contain deviations from a pure linear sweep. Figure 19 shows examples of FM demodulated time series. Rather than attempting to uniquely describe the attributes of each type of deviation, it is simpler to compute the spectrum of the FM demodulated data. The fundamental frequencies of chirp rates that have been observed have spanned 9 kHz to 170 kHz. Figure 20 shows a histogram of chirp rates observed near EWR and indicates that the most frequent rates have been 9, 26, 29, 72, 85, 118, 123, 159 and 170 kHz.
FIGURE 20. Histogram of EWR PPD chirp rates.
Examples of the 19 unique PPDs (detected within one month) are shown in Figure 22 through Figure 40, with the RF spectrum shown on the left, and the spectrum of the FM demodulated shown on the right. In each of these plots the scaling for the RF spectra is identical, spanning 40 MHz centered on 1575 MHz with a vertical scale using 10 dB per grid. All of the FM demodulated spectra use a horizontal axis that spans 0 to 200 kHz. For references purposes Figure 21 shows the RF spectra when no RFI is present.
Some of these PPDs were transmitting at power levels (observed by the PCTEL antenna) as much as 40 dB above the LNA noise floor. Most spectra were not centered symmetrically about L1 with some completely outside the mainlobe of the GPS C/A code. A few were transmitting outside the programmed 40 MHz bandwidth of the Snapshot System (1555 to 1595 MHz). For those PPDs transmitting outside this band, estimates were made of the upper or lower frequencies using the linear slope of the FM demodulated data, and then extrapolating that slope based on the chirp interval.
Some of the FM demodulated spectra contain a single spectral line that indicates the waveform modulating the RF has a very linear sweep. Most contain additional harmonic lines about the major component and a few appear to have bandwidth about their main spectral component. The current hypothesis is that many of these devices are poorly shielded and that the internal oscillator used to modulate the RF is affected by other nearby signals that are then appearing at the RF output. Some of the possible sources could be circuits that are within the device itself but there is some evidence that a few of these devices are susceptible to energy external to the PPD. A Snapshot System located near Houston International Airport (ZHU) has captured data from PPDs that contain strong components at 58.7 Hz in addition to its linearly swept 97 kHz waveform. Since this frequency is sufficiently different from utility AC power sources (60 ± 0.03 Hz), it has been hypothesized the vehicle carrying that PPD, also has a power inverter. Most power inverters are specified to provide a frequency output of 60 ± 3 Hz. Figure 32 shows a spectral notch that was present at that single frequency throughout the complete capture and suggests that that particular device may have had an impedance matching problem in its transmission path.
FIGURE 21. Spectra with No RFI Observed by PCTEL (click to enlarge).FIGURE 22. 1570 to 1583 MHz, Chirp 117.35 kHz.FIGURE 23. 1556 to 1583 MHz, Chirp 28.43 kHz.FIGURE 24. 1565 to 1578 MHz, Chirp 123.13 kHz.FIGURE 25. 1568 to 1583 MHz, Chirp 111.08 kHz.FIGURE 26. 1578 to 1589 MHz, Chirp 118.07 kHz.FIGURE 27. 1568 to 1584 MHz, Chirp 8.92 kHz.FIGURE 28. 1572 to 1584 MHz, Chirp 121.93 kHz.FIGURE 29. 1557 to 1622 MHz, Chirp 36.14 kHz.FIGURE 30. 1568 to 1582 MHz, Chirp 11.08 kHz.FIGURE 31. 1570 to 1585 MHz, Chirp 85.06 kHz.FIGURE 32. 1572 to 1582 MHz, Chirp 118.07 kHz.FIGURE 33. 1529 to 1577 MHz, Chirp 39.52 kHz.FIGURE 34. 1578 to 1594 MHz, Chirp 131.33 kHz.FIGURE 35. 1575 to 1582 MHz, Chirp 75.66 kHz.FIGURE 36. 1561 to 1586 MHz, Chirp 29.16 kHz.FIGURE 37. 1568 to 1592 MHz, Chirp 71.33 kHz.FIGURE 38. 1560 to 1595 MHz, Chirp 9.88 kHz.FIGURE 39. 1564 to 1582 MHz, Chirp 100.48 kHz.FIGURE 40. 1584 to 1599 MHz, Chirp 128.20 kHz.
A few snapshots have also provided evidence that some of the PPDs are erratic and probably not functioning as their manufacturer intended. Figure 41 contains a raster of 6 seconds of spectral data that shows a PPD whose output was meant to be between 1560 and 1580 MHz but for short periods of time was transmitting at frequencies above 1580.
FIGURE 41. Spectral Raster, PPD with Unstable Output.
Remarks on Additional PPDs
Since April 2011 the Snapshot System has captured many additional and different PPDs. No effort has been carried out to catalog all of the different types that have been observed but the following describes interesting and notable PPDs.
Figure 42 shows characteristics of a PPD that has had estimated EIRP approaching +20 dBm, to a point just outside the vehicle, and has also been associated with GPS receiver C/N0 degradation of more than -27 dB, strong enough to cause the WAAS receiver, located at the GBAS shelter, to lose lock on all GPS satellites for a short period of time.
FIGURE 42. 1568 to 1582 MHz, Chirp 118 kHz
Figure 43 shows a PPD that is one of the most frequently observed PPDs but that has not been associated with any significant degradation to the GPS receivers. Estimated EIRP for these PPDs has been on the order of no more than +10 dBm. One device with similar characteristics was procured and its measured power at the antenna output port was no more than +14 dBm. Marketing information on the internet for that PPD specified its output power as +25 dBm.
FIGURE 43. 1572 to 1589 MHz, Chirp 85 kHz.
Figure 44 shows a PPD that is transmitting at both L1 and L2. The EWR Snapshot System has been configured to only capture snapshot data at L1 due to the fact that LAAS only uses L1. However, the WAAS receiver used to monitor for RFI is a dual frequency receiver that on occasion has indicated simultaneous RFI at both L1 and L2. Even though the EWR Snapshot System has not captured data at L2, the simultaneous presence of both L1 and L2 RFI, provides strong circumstantial evidence that this RFI source was transmitting on both frequencies. A Snapshot System monitoring the WAAS Reference Station (WRS) at Leesburg Virginia has captured simultaneous L1 and L2 RFI events. Demodulation of that data indicated the two RF outputs had similar modulation but the demodulated data was not coherent. Therefore, that PPD was probably using individual, but similar waveform generators, for each RF output.
FIGURE 44. 1562 to 1583 MHz, Chirp 114 kHz.
Almost all PPDs have been observed individually. However, there have been at least three times in the last two years when two unique PPDs have been observed within 60 seconds of each other. Figure 45 plots normalized degradation in C/N0 while Figure 46 plots snapshot measured power for the same RFI event. Analysis of snapshot data for each of the times that had strong RFI power are shown in Figure 47 and Figure 48 and confirmed that there were in fact two unique PPDs observed approximately 40 seconds apart. Both were traveling south on the NJT and approximately 1200 meters apart.
FIGURE 45. Normalized C/N0 August 19, 2011FIGURE 46. Snapshot Power August 19, 2011.FIGURE 47. C/N0 -19.0 dB, Chirp Rate 78.97 kHz.FIGURE 48. C/No -28.0 dB, Chirp Rate 117.24 kHz.
Most of the observed RFI events last for no more than 50 seconds although a few that lasted much longer have been correlated with slow traffic on the NJT. Figure 49 is from June of 2010, before the updated monitoring equipment was in place, and displays normalized degradation of C/N0. The time duration for which this RFI was observed was more than 3 minutes and was during a time when traffic was ‘slow’ on the NJT.
FIGURE 49. June 9, 2010, PPD, Estimated Velocity 10 m/s.
Very wide-bandwidth PPDs have recently been observed more often. The frequency span these devices are transmitting has had to be estimated due to the fact that the Snapshot monitor has 40 MHz of bandwidth, and these PPDs are transmitting beyond this bandwidth. Figure 50 through Figure 52 show examples of these types of PPDs. The left plot in these figures is the RF spectra and the right plot is the FM demodulated waveform. The latter each contain a linear component that is present for only a portion of the chirp interval. Under the assumption that the modulating waveform would be linear for the repetition interval, the slope of the visible linear component was extrapolated to the total chirp time interval. It is not possible to estimate the upper and lower frequency points for the last two examples, since neither of those had a frequency that began or ended within the observable 40 MHz bandwidth of the monitor.
Although the Snapshot System L-band tuners can be programmed for greater bandwidth, the limiting bandwidth is the bandpass filters contained within the LNA modules, which have bandwidths of 40 MHz.
Data captured by a Snapshot System operating near ZHU contains evidence that external energy may have coupled into that PPD and affected the modulation waveform. Figure 53 shows a plot of the RF spectra and an expanded portion of the FM demodulated spectra indicating the presence of a 58.7 Hz component. A raster of the demodulated FM, shown in Figure 54, highlights the 58.7 Hertz component.
FIGURE 53A. ZHU chirp 118 kHz with 58.7 Hz.FIGURE 53B. ZHU chirp 118 kHz with 58.7 Hz.FIGURE 54. ZHU raster of FM showing 58.7 Hz (Click to enlarge).
Careful analysis of normalized C/N0 has also provided clues as to the possible travel paths that a PPD might be using. RFI was suspected at the WRS located at Leesburg Virginia (ZDC). A Snapshot System was installed to detect and characterize possible RFI. Analysis of snapshot data did confirm that a few PPDs were traveling past ZDC. One of the PPDs was more disruptive than the others but fortunately was also following a very predictable schedule. It was regularly detected twice a day, first within 10 minutes of 4:30 AM local and next within 30 minutes of 2:30 PM. Normalized C/N0 contained similar patterns for each time of day and are shown in Figure 55 and Figure 56. Examination of the local roadways, shown in Figure 57, suggested the possible roads and direction in which this PPD was traveling. The WAAS antennas on the roof of ZDC have clear line of sight to state highway 7 for vehicles that are east of ZDC. Normalized C/N0 for morning events tended to have a relatively abrupt onset followed by a gradual return to normal while the afternoon events exhibited a gradual increase in degraded C/N0 followed by a quick return to normal. This observation lead to hypothesizing that the PPD was traveling east in the morning and west in the afternoon.
FIGURE 55. ZDC Typical Morning Degradation.FIGURE 56. ZDC Typical Afternoon Degradation.FIGURE 57. Roads near ZDC (click to enlarge.)
FAA Spectrum personnel were informed of this analysis and confirmed that this hypothesis was correct. Using this information they were able to detect the vehicle that was responsible and remove this particular PPD from service.
EWR RFI Event Statistics
A large number of RFI events have been detected at EWR since the updated Snapshot System was installed on March 3, 2011. These RFI events have been ranked according to the magnitude of degradation in normalized C/N0, as reported by the GBAS shelter WAAS receiver. The following plots show the total number of RFI events per day (red dots) and for every seven consecutive days (blue line). On average, PPD-induced receiver degradation of at least 10 dB has been observed two times a day. Although a small number of narrowband RFI events produced receiver degradation of as much as 10 dB, the vast majority of RFI events causing 10 dB or more of receiver degradation are due to PPDs.
Figure 58 indicates that since March 2011, more PPD RFI events are being observed. However, Figure 60 indicates that the higher-power PPDs are not being observed as often. One possible explanation is that the previously observed PPDs have stopped working, and the individuals using them have either not acquired replacements or they have acquired different ones with less-damaging RFI.
FIGURE 58. History of RFI > –10 dB (click to enlarge).FIGURE 59. History of RFI > –15 dB (click to enlarge).FIGURE 60. History of RFI > –20 dB.
Many recent PPDs have been transmitting with estimated frequency spans of 65 MHz to 140 MHz. Although the estimated EIRP of many of these very wide bandwidth PPDs has been as much as +10 dBm, their effect on GPS receiver processing has not been as damaging due to the fact that the RFI is within the GPS receiver processing bandwidth for only a portion of the time.
Spectra of Swept FM with Multipath
One of the most commonly observed characteristics in PPD spectral data has been uniformly spaced nulls as shown in Figure 61 and Figure 62. Figure 63 displays a spectral raster that shows how the nulls shift in frequency over time. Initially, there was uncertainty as to the mechanism responsible for these observations. Under the hypothesis that multipath might be responsible, a single ray multipath model was used to predict the spectral characteristics of a PPD that includes a multipath component. This analysis was pursued in the hope that it might provide additional information as to the exact location of PPDs.
The swept CW signals used by PPDs provide a useful source for characterizing multipath between the PPD and monitoring antennas. Equation (1) models observed CW that is the sum of the direct path and a single reflection in which reflected component has a path length difference of d meters.
(1)
Assuming the reflection is from a metal surface, it should experience a phase reversal. Therefore, destructive cancellation between the direct and multipath component will be present for those frequencies that have a path length difference that is an integer multiple of the wavelength. Equation (2) represents this condition.
(2)
Simplification results in the following expression.
(3)
As an example, if spectral nulls are observed at intervals of 10 MHz, then the path length difference is approximately 30 meters. Spectral nulls have been observed at frequency intervals ranging from 2 MHz to as much as 30 MHz. These null spacing’s translate to path length differences of between 150 meters to 10 meters. Multipath with a path length difference of less than 8 meters will exhibit a single null in the 40 MHz bandwidth of the Snapshot System and therefore cannot be estimated accurately using this technique. A path length difference of 8 meters is also what might be expected for two vehicles traveling side by side on interstate highways since interstate highway specifications require lanes to be approximately 4 meters wide.
Once a possible mechanism for the spectral nulls was hypothesized, additional analysis was performed on specific RFI events in which uniformly spaced spectral had been observed. Snapshot and GPS receiver data indicates the direction of travel for a PPD. With direction of travel known, it is possible to approximate the distance that the PPD is from the monitoring equipment. However, for those RFI events that were examined, the calculated path length difference was similar to or greater than the distance between the PPD and monitoring equipment. The most likely location of surfaces that would reflect the PPD transmission was other vehicles on the NJT. Had the surfaces responsible for the reflections been stationary objects nearby, then it might have been possible to hypothesize the most likely location of the PPD by combining receiver proximity and path length differences.
The magnitude of the reflection coefficient can be estimated by comparing the relative power of the spectral maximum and minimum. However the magnitude of the reflection coefficient can only be bounded since it depends on both the reflection coefficient and the relative path length difference. Since the reflected path travels farther, its magnitude will inherently be reduced, in addition to the loss from the reflection, and therefore the observed relative difference will be smaller than shown by equation (4).
(4)
For the examples shown in Figure 61 and Figure 62 the spectral max/min was on the order of 10 dB. By using 10 dB for SpectraMaxMin in equation (4), a reflection coefficient of at least 0.5 is calculated. Reflection coefficients of trucks with shipping containers will probably be much greater than 0.5 and could easily be as high as 0.9.
Direction-Finding Methods
After examining more than a thousand examples of PPDs and their effect on GPS receivers, I have concluded that any type of ground-based direction finding system intended to detect and locate low power moving PPDs over a large area using time-difference-of-arrival (TDOA) or beam forming (angle-of-arrival, AOA) techniques will face significant challenges.
Accuracy of TDOA-based location systems can be decomposed into two components: measurement accuracy, and the geometry of the equipment used to make these measurements. In 1982, Paul Chestnut calculated the relationship between accuracy and these two components, showing that geometry is a multiplicative factor. Direction-finding conceptual design typically strives to position the measurement equipment such that it surrounds the area to be monitored, if possible. For those situations where it is not possible to encircle the area, the measurement equipment will typically have a long baseline between its sensors and with a perpendicular orientation with respect to the monitored area. This strategy reduces errors due to geometry.
Measurement error depends on the observable power of the signal to be located. This component will most likely limit the ability to accurately locate low-power moving PPDs. Measurement data from GPS receivers demonstrate the ability to reliably detect the presence of these PPDs, but only when they have been within hundreds of meters. Most PPDs have been observed for 30 to at most 60 seconds. For PPDs traveling along the NJT at a velocity of 30 m/s, this time span implies that they were not detected until they were within 900 meters. Reliable detection was not demonstrated unless they were within 500 meters. This is a direct consequence of the fact that all GPS antenna are designed for hemispherical coverage for satellites above the horizon. The PPDs are at the horizon or lower relative to most GPS antenna patterns. If GPS antennas were used as part of a TDOA-based system, they would need to be positioned no further apart than approximately 600 meters. Using GPS antennas to detect and locate PPDs is inherently limited to proximity detection.
The ability to locate low-power PPDs within a larger area requires a system to be able to detect a PPD by all of its sensors simultaneously from distances much greater than 600 meters. In principal, higher-gain antennas orienting their main beam along the NJT would increase the distance over which PPDs could be detected. Even when high-gain directional antennas have been positioned along the NJT, the observed power has not followed predictions based strictly on FSL. One reason for this is that clear line-of-sight could not be achieved between the high gain antenna and the PPD. Vehicular effects might also be responsible for these observations. What is known is that the power observed by ground-based antennas has shown significant fluctuation, much greater than could be explained by FSL propagation. Ground-based direction finding systems for low-power moving PPDs must be able to simultaneously detect RFI by multiple sensors, in an environment that may at times shield the PPD from its sensors, and that also has fluctuating observable power.
The ability to observe, detect, and locate low-power moving PPDs over a large area would require a system to have its monitoring antenna located high above the NJT, and oriented to look down on the NJT and the surrounding area. Although such a system must still contend with shielding from the vehicle itself, it would be the only approach that could potentially observe these PPDs over a large area, simultaneously by all of its measurement sensors. Implementing such a concept may be expensive.
A less expensive approach might rely on proximity detection, with many sensors on the ground, each monitoring their own small area, and only report PPDs that travel close to each of them. One such approach, referred to as crowd sourcing, uses cell phones to aid in detecting and locating PPDs. Crowd sourcing is a form of proximity detection.
Summary
The first evidence of low-power moving PPDs along the NJT used two GPS receivers separated by considerable distance and then correlated their responses. This approach is only useful for detection of PPDs traveling in close proximity to the GPS receivers. Analysis of GPS receiver normalized C/N0 also provided a basis for determining if RFI might be from a moving emitter. The shape of normalized C/N0 versus time not only provides clues that the RFI source is in motion, but may even be correlated with their possible travel paths, when blockage exists between the GPS antenna and local roadways.
Autonomous operation is a necessity for detecting low-power moving PPDs, since they may be observable only a few times a day and for less than 60 seconds.
Capturing real-time samples of these intermittent RFI events determined the existence of many different types of PPDs. Almost all use some form of swept FM modulation. Analysis of their spectra and modulation indicates that many devices are probably not operating as their manufacturer intended. Modulation waveforms of PPDs have included triangle, highly linear sawtooth, sawtooth with synchronous perturbations on top of the fundamental sawtooth, and sawtooth with analog modulation. The leading hypothesis for this observation is that the devices are not shielded very well and that the internal modulator is susceptible to coupling from either other circuits within the PPD itself or from external electronic devices operating in the vicinity of the PPD.
Acknowledgments
The author thanks the FAA for support in this investigation, Rich Holley of ICE-Online for help in utilizing capabilities of the ICEPOD6-M5, and Julian Babel of the FAA Technical Center for regularly swapping external hard drives attached to the Snapshot System.
Joe Grabowski is a systems engineer at Zeta Associates where he works on a variety of GPS projects in support of the FAA, as well as communications systems and digital signal processing applications. He received an M.S.EE from Purdue University. Since 2010 he has been involved in the investigation of personal privacy device impacts on FAA SBAS and GBAS sites.
Non-aviation users of satellite- and ground-based augmentation systems do not require the conservative level of integrity built into these systems for aviation users. Removing it can produce substantial benefits in terms of smaller error bounds and improved availability.
By Sam Pullen, Todd Walter, and Per Enge
Both space-based and ground-based augmentation systems (SBAS and GBAS, respectively) are designed to enhance standalone GNSS navigation to meet the requirements of civil aviation. SBAS and GBAS corrections and integrity information are also available to the non-aviation user population, such as automobiles, buses, and trains on land as well as ships near shore. This much larger user base can benefit as much from the integrity components of SBAS and GBAS as from the increased accuracy obtained from applying SBAS and GBAS pseudorange corrections. However, there are significant differences between the aviation interpretation of navigation integrity and the interpretation that would be natural to most users.
SBAS and GBAS provide integrity in a multi-step procedure that is laid out in the RTCA Minimum Operational Performance Standards (MOPS) for the FAA versions of both systems: DO-229D for the Wide Area Augmentation System (WAAS) and DO-253C for the Local Area Augmentation System (LAAS). These systems indicate which ranging measurements should be excluded as unsafe to use and provide bounding error standard deviations, or sigmas, for the remaining usable measurements. Each aircraft uses this information to compute vertical and horizontal protection levels that define position-domain error bounds at desired probabilities. This process is straightforward, logical, and is not limited to aviation users. However, the requirements and assumptions underlying it make it very conservative.
SBAS and GBAS are designed to meet integrity requirements defined in terms of what is known as specific risk. Briefly, this means that all safety requirements must be met for the worst combination of knowable or potentially foreseeable circumstances under which an operation may be conducted. Some variable factors important to safety, such as the user’s satellite geometry, are known by definition. Others, such as receiver thermal noise, are random and unpredictable. But several factors that are critical to GNSS performance, such as multipath and ionospheric errors, are neither completely random nor deterministic. Specific risk typically treats all error sources that are not completely random in a worst-case manner. SBAS and GBAS are designed to mitigate specific risk to support civil aviation, and the resulting conservatism makes SBAS and GBAS less attractive to non-aviation users who expect tighter protection levels relative to nominal system accuracy.
Fortunately, non-aviation users need not apply all MOPS procedures required of aviation users if their own safety requirements differ. Most users define integrity in average or ensemble terms, meaning that everything not known in practice is treated as random and is probabilistically mixed (or convolved) together. The protection levels valid for these users would be much lower than for aviation users, even though the stated bounding probability is the same. This contrast is illustrated in Figure 1, which shows example bounds on 2-D vertical errors at a probability of 0.95 (the 95th percentile, or 95 percent) for accuracy and a probability of 1–10-7 for integrity. The term VPE stands for vertical position error, while VPL stands for vertical protection level. Analogous terms (HPE and HPL) and a similar picture exist in two dimensions for horizontal errors.
Only one 95 percent error bound is shown in Figure 1 because this probability can be observed, estimated, and modeled with theory and reasonable amounts of data (hundreds or thousands of independent samples). This is not at all the case at the very small probability of 10-7 that applies to aviation precision approach: it is roughly equivalent to one event in 47.5 years per 150-second precision-approach interval. Both theory and data fall far short of being able to predict such rare-event errors. Extrapolating from available data to 1–10-7 using Gaussian distributions is perilous because the Gaussian distribution almost never applies at such small probabilities. Mixed-Gaussian models, other so-called fat-tailed distributions, and inflation of Gaussian parameters help address this, but the uncertainty regarding the true error distribution results in significantly different error bounds depending on the assumptions that are made. The same is true regarding the effects of faults and anomalies that are more probable than 10-7 but are still rare and poorly understood.
In the end, different means of assessing these uncertainties and various degrees of user risk aversion result in different 1–10-7 protection levels, as shown in Figure 1. It is this difference that we wish to quantify and exploit in this article.
Average versus Specific Risk
The concept of average or ensemble risk is intuitive to those with a background in probability and is one of the key principles of probabilistic risk assessment (PRA). Thus, it helps to examine it first.
Average risk is the probability of unsafe conditions based upon the convolved (averaged) estimated probabilities of all unknown events. More specifically, probability distributions are derived (based on the best available knowledge) for all unknown parameters relevant to user safety, and these are combined (by probabilistic convolution) to create an overall distribution that represents safety risk as a function of the known parameters. This straightforward, natural interpretation of probability and uncertainty has a major advantage in that it cleanly separates the probabilistic calculation of safety risk from users’ aversion to risk. By keeping risk probability and risk aversion (or severity) separate, a final risk consequence measure can be derived that supports apples-to-apples comparisons of alternatives. One useful result of this is known as the value of information (VOI). By comparing the risk outcomes of two scenarios in which the latter case has additional information (for example, from an additional sensor or integrity monitor), the risk-reduction benefit of the added information can be traded off against the cost and complexity that it introduces to the system. Similar comparisons can be made for any definition of risk, but the definition and use of VOI in an average-risk framework makes the most sense in both theory and practice.
Turning to specific risk, no single definition exists within the aviation safety community, to our knowledge. This is partially because of the uniqueness and complexity of the concept and partially because multiple inconsistent interpretations appear to exist. Therefore, we provide our own definition: Specific risk is the probability of unsafe conditions subject to the assumption that all credible unknown events that could be known occur with a probability of one (on a risk-by-risk basis).
To understand how specific risk differs from average risk, it helps to start with a fault-tree representation of risk in which loss of integrity (LOI) can result from any of the nodes of the tree. Figure 2 shows a simplified example of a fault tree for CAT I GBAS. It shows the allocation of the CAT I total integrity risk requirement of 2 × 10-7 per approach to the various possible causes of integrity loss. In specific-risk analysis, each type of failure shown in the tree, if deemed to be a credible failure (meaning, in practice, that its assumed prior probability is larger than compared to its allocation in the fault tree), is assessed that the failure is guaranteed to occur in a worst-case fashion. This means that the variables that describe this particular failure scenario take the values that maximize the hazard to users. In an average-risk analysis, these variables would take many values according to their own probability distributions, and these distributions would be convolved together to provide an overall representation of risk under that scenario. Instead, one scenario drives the specific risk assessment for a particular user class, and it is the worst one possible from that user’s standpoint. (Another user class would be evaluated under a different set of parameters corresponding to the separate worst case for that user.) The improbability of the worst-case combination of parameters is not considered as long as the probability of the failure scenario as a whole is deemed high enough to be of concern.
Figure 2. Fault tree for CAT I GBAS integrity.
Since GNSS augmentation systems contain multiple levels of health monitoring, the worst-case scenario is usually the one that maximizes the probability of an undetected hazardous error for a particular user class. Hazardous error is typically defined as any error that exceeds a pre-defined safety zone known as an alert limit (AL) or any error that exceeds the computed protection level (PL), which allows integrity to be defined separately from the intended application. Both definitions are conservative in that all errors exceeding AL or PL are treated as equally hazardous. In other words, an error just above AL is treated as just as dangerous as an error of 10 × AL. They are also misleading when used in specific-risk analyses because the resulting worst-case conditions are those that give errors just above AL or PL, as these are the generally hardest for monitoring algorithms to detect.
The use of specific risk in aviation is an evolution of deterministic guidelines for tolerable risk that date back to an earlier era when flying was more dangerous. It remains dominant in aviation safety assessment because it is partly responsible for the development of safer and more reliable air transportation. However, it has several important weaknesses compared to average risk. The first is that the degree of risk aversion preferred for aviation is buried within the hazard probabilities generated by specific risk — it cannot be separated out. This means that specific-risk results do not translate well to other classes of users, as very few users would happen to have the same risk preferences that have evolved within aviation over several decades. In addition, specific risk makes a distinction between unknown events that could be known and those that are both rare and completely unknowable. A very risk-averse value of information is much different than the risk-neutral one built into PRA, as it severely penalizes systems that do not include all potentially-informative sensors. Since each sensor added to a system provides less benefit than the last, almost all cost-effective systems choose to include less than the maximum possible number of sensors.
The conservatism implicit in specific-risk assessment severely penalizes users. Although PRA would show that the combination of factors (shown in an example induced by extreme ionospheric spatial decorrelation) needed to produce a 40-meter error in a CAT I GBAS system is exceedingly improbable (almost certainly below 10-10 per approach), specific risk forces a significant part of the GBAS risk-mitigation effort to be targeted at this scenario. In this case, since monitoring is not guaranteed to detect the anomaly in time, the only recourse is geometry screening, a cumbersome technique in which the ground system continually evaluates the worst-case error and, if it exceeds a 28-meter tolerable limit at the CAT I decision height, determines which broadcast parameters to inflate such that all satellite geometries causing worst-case errors exceeding 28 meters are made unavailable (because the inflated VPL is larger than the 10-meter CAT I VAL). The result of this procedure is much lower user availability than would be achieved without inflation. SBAS pays a similar penalty, as we will see later. The broadcast grid ionospheric vertical error values that bound worst-case ionospheric errors (and thus the resulting protection levels) are much higher than they would be if the unusual combination of factors needed to create the worst-case error scenario were not the dominant concern.
To the extent that loss of availability represents a safety issue at the airspace level, the worst-case focus that results from specific risk is not optimal even from a safety standpoint. But this is not the only concern. Specific risk requires a great deal of development and testing to identify and mitigate a handful of very peculiar, non-representative conditions. When schedule and resources are limited, other potential threats that are easier to foresee but seem extremely improbable are often neglected. One example is the treatment of multiple hardware failures. If individual failures are assumed to be statistically independent, the probability of multiple simultaneous failures is very small. However, while statistical independence is a common assumption in probability classes because it makes calculations easier, it rarely applies in the real world. Because satellites and ground receivers are similar, if not identical, the presence of a failure in one unit may suggest a common cause or at least a common vulnerability, meaning that the probability of additional failures is much higher than independence would suggest. Thus, assuming independence by default could lead to neglecting entire categories of risk that are more threatening than the worst-case events that dominate specific risk.
Maximum WAAS Errors, Protection
To investigate the conservatism built into SBAS and GBAS specific risk assessment, maximum WAAS horizontal and vertical position errors over time (as measured by the Performance Analysis Network (PAN) maintained by the William J. Hughes FAA Technical Center) have been examined and compared to the protection levels computed when the maximum errors occurred. This study begins with PAN Report #8 (covering January to March 2004 — shortly after WAAS commissioning in mid-2003) and extends through PAN Report #34 (covering July to September 2010). Each PAN report covers three months of observed WAAS performance.
Figure 3 shows the 38 WAAS reference stations (WRSs) used by the PAN to collect position error and protection level information (some of these stations were not active in 2004 and thus were not used in earlier PAN reports). While measurements from these stations are used to generate WAAS corrections and error bounds, they are also used by the PAN as static pseudo-users that compute WAAS-corrected positions and protection levels according to the aircraft user algorithms specified in the WAAS MOPS. The resulting positions are compared to the known, pre-surveyed positions of each station to derive estimates of vertical and horizontal position errors (VPE and HPE) once per second.
Figure 3. WAAS PAN reference station network.
Figure 3 groups these stations into three sets of stations based on their presumed quality of WAAS coverage. These sets are unofficial and were created for the purposes of this study. The seven stations in the inner set are expected to have good WAAS coverage at all times because they are surrounded by other stations. The 13 stations in the outer set are expected to only have acceptable coverage because s
ome of them are at the edges of CONUS. The remote stations provide coverage to the inner and outer regions as well as the best possible coverage of their own regions. Because the remote stations extend beyond the primary coverage region of WAAS in CONUS, errors at these stations are not considered here.
Figure 4 is a 2-D plot of position error versus protection level in the vertical axis (that is, VPE versus VPL) for all epochs and stations during the three months covered by the recent WAAS PAN Report #34 (July 1–September 30, 2010). These results are typical of the entire period since WAAS commissioning in 2003, particularly the last several years. The vertical lines on the plot indicate the 95th-percentile, 99th percentile, and maximum VPEs in this period (1.2, 1.8, and 7 meters, respectively). The maximum VPE occurred at Barrow, AK, which is one of the most remote stations in the WAAS network (see Figure 3). In comparison, the lowest VPLs (intended to be 1–10-7 bounds on VPE) are in the range of 10–15 meters, and values as high as 40 meters are not uncommon. The most demanding approach operation that WAAS supports, LPV-200, allows approaches to a 200-foot minimum decision height and requires that VPL be below a vertical alert limit (VAL) of 35 meters. HPL must also be below a horizontal alert limit (HAL) of 45 meters. When this is not the case, the approach operation is not available; thus these higher VPLs extract a significant cost.
Figure 4. WAAS vertical protection level versus vertical position error (June–September 2010).
Figure 5 and Figure 6 (for vertical and horizontal errors, respectively) span the entire period of WAAS PAN Reports used in this study. VPL represents the VPL at the station and time of the maximum VPE; it is not the largest VPL recorded at a particular station. The horizontal errors shown in Figure 6 are defined analogously. Note that the station that observes the largest horizontal error in a given PAN report may differ from the one that observes the largest vertical error.
Figures 5 and 6 demonstrate that, while both 95 percent and maximum errors are quite low and are within the expected range of each other, the protection levels associated with the maximum errors greatly exceed them. This pattern is clearer in Figure 5 for vertical errors because maximum VPL tends to be more consistent across PAN reports, but it is true for horizontal errors as well.
Figure 5. WAAS vertical errors and protection levels from 2004–2010.Figure 6. WAAS horizontal errors and protection levels from 2004–2010.
Figures 7 and 8 clarify this relationship by plotting the ratio of VPL to VPE and HPL to HPE for the station and time of the maximum error. The mean of this ratio is very high and is about the same in both cases: 5.38 for vertical and 5.21 for horizontal. Figure 7 shows a steady upward trend in the ratio that is mostly due to WRS improvements that resulted in maximum VPE being reduced over time. This trend is clearly visible in Figure 5 and appears to exceed the weaker trend of lowering VPL due to WAAS algorithm enhancements. The same trend is visible in the horizontal Figures 6 and 8 but is weaker due to the greater variability of maximum HPL over time.
To evaluate the significance of the large PL-to-max-PE ratios in the WAAS PAN database, we need to approximate the number of independent samples from which the maximum errors were derived. As noted before, WAAS protection levels represent error bounds at the 1–10-7 probability level based on specific risk. With one measurement being collected at each operational station every second, a total of about 4.25 billion samples were collected in the PAN reports from January 2004 to September 2010. Note that measurements from remote stations are included in this count, but they are also represented in the conclusions because their PL-to-max-PE ratios are very similar to the ones shown in Figures 7 and 8. Translating this number into the number of statistically independent samples depends on the interval between independent measurements. Because both nominal and rare-event errors affect this interval, it is hard to estimate. Our best guess is a range between roughly 30 and 150 seconds, suggesting that the PAN database contains between 2.8 × 107 and 1.4 × 108 independent samples. Both of these numbers suggest that WAAS protection levels are very conservative from the perspective of average risk.
Figure 7. Ratio of VPL to VPE from 2004–2010.Figure 8. Ratio of HPL to HPE from 2004–2010.
Adjusting for Average-Risk Users
Using the above results, a preliminary estimate of the reduced WAAS protection levels that would apply to average-risk users can be made. Figure 9 shows a comparison between the actual 95 percent WAAS VPL and HPL and the adjusted VPL and HPL potentially achievable with WAAS (for the same 1–10-7 bounding probability) for average-risk users. The actual WAAS VPLs are taken from the more recent WAAS PAN Reports starting from #24 (covering January to March 2008) as the period from 2008 to 2010 includes most of the WAAS algorithm improvements introduced since commissioning in 2003. The actual 95 percent VPLs and HPLs represent the largest reported 95th-percentile values among the stations within CONUS for each quarterly period. The lower adjusted VPLs and HPLs are derived by dividing each VPL by a factor of 4.0 and each HPL by a factor of 2.5. These two reduction factors are derived from Figures 7 and 8, respectively, as conservative estimates of the ratio between protection levels and maximum position errors. Note that the factor of 2.5 for horizontal errors does not include the 12-meter error in Cleveland from PAN Report #13, as this is thought to be spurious (that is, not representative of actual WAAS behavior).
Figure 9. Projected WAAS protection level reductions for average-risk users.
While projections based on these reduction factors are imprecise, they demonstrate the much lower error bounds that non-aviation users with an average-risk safety perspective could achieve. Most non-aviation users operate on land or sea and will be primarily concerned with horizontal error bounds. Figure 9 suggests that the typical 95th percentile WAAS HPLs of 15–20 meters (for the worst location in CONUS) can be reduced to 6–8 meters and still provide a confident 1–10-7 error bound.
It is important to emphasize that these preliminary projections for average-risk users are just that. In order to formally establish new integrity requirements and protection levels for existing systems, the hazardously misleading information (HMI) analyses previously done for these systems need to be redone using the principles of PRA and average risk. While the original development of the WAAS and LAAS HMI analyses was lengthy and resource-intensive, almost all of the detailed work is already complete. As long as the original analyses are available, it is a much smaller task to take these results and create PRAs out of them by extracting the original specific-risk assumptions and applying average-risk principles instead.
LAAS Users. Since the first GBAS ground station design (the Honeywell SLS-4000 LAAS Ground Facility) was certified for CAT I use in 2009 and has not yet been approved for operations at a specific airport, much less data is available to do a preliminary analysis for GBAS similar to the one done for WAAS above. However, the degree of sigma inflation in the parameters broadcast by CAT I LAAS is approximately known, meaning that it can be more-precisely removed from the current LAAS protection levels to estimate what they would be for average-risk users.
Figure 10 shows the degree of inflation applied to the broadcast σvertical_iono_gradient (or σvig) parameter in order to protect against the worst-case ionospheric anomaly described previously. This result is for the SPS-standard 24-satellite constellation over a 24-hour period at the LAAS installation at Newark Airport, New Jersey (the method used by the Honeywell SLS-4000 is somewhat different). While not all epochs require inflation, a majority cause the nominal σvig value to be increased by a factor of 2 or more, which significantly decreases CAT I availability and currently makes it impossible to take advantage of the Differentially Corrected Positioning Service (DCPS) for non-CAT-I operations.
Figure 10. Typical σvig inflation factors for CAT I LAAS.
Because of the extreme rarity of the worst-case event that dictates this inflation, it would likely not be needed for average-risk users. Figure 11 shows how much the σvig inflation in Figure 10 increases the LAAS VPL at Newark for the standard 24-satellite constellation. The VPL reduction from removing the inflation is not as dramatic as the potential reductions shown for WAAS in Figure 9, but they are significant relative to the 10-meter VAL for LAAS CAT I approaches. Furthermore, the pre-inflated nominal value of σvig for LAAS is 6.4 millimeters/kilometer, which is much higher than the actual one-sigma nominal gradient value of 1–2 mm/km because, under specific risk, the very worst nominal data must be bounded (also, worst-case tropospheric gradients must also be bounded by σvig). Other broadcast parameters that affect VPL, such as σpr_gnd and the ephemeris P-value that bounds worst-case ephemeris failures, would also be reduced significantly by switching to average risk. Overall, it is likely that LAAS protection levels based on average risk would be reduced from the current specific-risk PLs by about the same range of factors (2–5) observed from WAAS data.
Figure 11. Impact of σvig inflation on LAAS VPL.
User Performance Improvements
This discussion assumes that most non-aviation users who are not encumbered by the history of aviation standards development will prefer to quantify risk using PRA and the average-risk approach. As noted earlier, average risk better matches most users’ intuitive understanding of uncertainty and has the enormous advantage of separating risk quantification from risk aversion. Regardless of how risk-averse or conservative a given operator is, his or her model of risk aversion can be applied most efficiently to a risk-neutral calculation of risk that fairly represents all aspects of uncertainty. Inserting risk aversion into the calculation of risk, as done in the specific-risk approach, is both inefficient and non-optimal from a safety perspective because extensive focus on a few extreme worst-case events drives attention away from other events.
The HPL reductions for average-risk users illustrated here would be significant for many classes of ground and marine transportation users. They would allow operations with tighter physical safety margins to be supported. Users who gain no particular benefit from tighter protection levels would still obtain much higher availability of integrity, as a 25-meter HPL could be supported by much poorer satellite geometries than would otherwise be the case. In other words, users that can tolerate 25-meter horizontal error bounds would be able to operate safely a much higher percentage of the time, because the degree of GNSS constellation deterioration needed to exceed this limit would occur much less often. These benefits do not only apply at the 1–10-7 probability level, as they would scale to the higher probabilities (1–10-4 to 1–10-6) that many non-aviation applications would be most concerned with.
While very few non-aviation users of GNSS today have real-time safety requirements similar to those of civil aviation, the number of such users will likely increase as the coverage of augmented GNSS (and the availability of integrity from standalone receiver-autonomous integrity monitoring, or RAIM) expands. The evolution of standalone civil GPS usage provides a precedent: as basic GPS accuracy improved from tens of meters to several meters, and the cost of user equipment dropped, more and more uses were discovered. A similar, although smaller-scale, trend is likely to occur as the advantages of augmented GNSS become more available and better understood. The primary beneficiaries are likely to be intelligent road-transport systems, train services, and marine transportation in restricted waters.
One application where tight real-time integrity bounds would be useful is in harbor and marina entry and exit; see Figure 12, taken from a Google map of a marina in San Diego, California. Based on the earlier analysis, two typical 1−10-7 horizontal protection levels are shown: 18 meters using the unchanged WAAS MOPS approach, and 7 meters based upon modifying the broadcast bounding parameters to represent average risk (these HPLs are bounds on error in either direction, positive or negative; thus the 2-D error bounding circle has a diameter of twice the HPL).
Figure 12. Example of reduced protection levels for harbor/marina access.
When the resulting error bounds are compared, the relative advantage of the smaller bound for this application is immediately apparent. In general, when HPL is significant compared to potential obstacles, its significance varies with the square of HPL rather than HPL itself, as the area being protected matters more than either linear direction. In this example, the ratio of HPLs being compared is 18/7, or 2.57, but the ratio of HPL-squared is much larger: 182/72 = 6.61.
When real-time integrity is not needed, augmented GNSS provides an easy means to guarantee or certify vehicle locations after the fact with great precision and reliability, without the need for post-processing. Vehicle and cargo tracking based on standalone GPS is common today, a certification of the correctness of the tracking data to probabilities suitable for legal or commercial guarantees is lacking. For this, error bounds at 1–10-4 to 1– 10-6 probabilities are likely sufficient, and would allow HPLs of below 5 meters from WAAS and below 3 meters from LAAS. In some scenarios, the difference between a 5-meter and a 15-meter guarantee would be minor, but in others, it could make a substantial difference.
As noted earlier, even for uses where the required HPL (as represented by the safe error limit, or HAL, for a particular application) is satisfied by the existing WAAS and LAAS protection levels, the use of modified average-risk protection levels increases the availability of integrity, which is most often expressed as the probability or percentage of time (over all satellite geometries and othe
r variable system states) that the integrity requirement is met throughout an operation (in simple terms, that HPL ≤ HAL). For user locations within good WAAS or LAAS coverage, the most variable element over time is satellite geometry. Decreasing HPL by a factor of 2.5 or more substantially increases the margin between HPL and HAL and makes it far less likely that the satellite geometry will degrade to the point where HPL exceeds HAL. For example, if the unmodified WAAS HPL equals HAL at an (un-weighted) HDOP of about 1.5, the resulting satellite availability (an upper bound on overall availability) for the SPS-standard 24-satellite GPS constellation would be roughly 98.5 percent. This means that the satellites in view (in this case, all satellites above 5 degrees elevation at a location in CONUS) would provide HDOP ≤ 1.5 about 98.5 percent of the time. However, the modified average-risk HPL (using the factor-of-2.5 reduction) would roughly translate into a limiting HDOP of about 3.75. This allows the required integrity bound to be satisfied by much poorer GPS geometries and gives a satellite availability of greater than 99.9 percent. Thus, when integrity is needed, this much greater availability of integrity is a major advantage.
Summary
SBAS and GBAS broadcasts are freely available to all GNSS users, most of whom will have different definitions of acceptable risk. These users are not optimally served at present and may hesitate to take advantage of SBAS and GBAS as a result.
Using years of collected data for the FAA WAAS system and analysis of the inflation factors built into the CAT I version of the FAA LAAS system, it appears that average-risk users of WAAS and LAAS would be adequately supported by protection levels that are 2 to 5 times lower than those currently derived by aviation users. The fact that two different approaches used to examine WAAS and LAAS suggest similar levels of over-conservatism lends credence to these estimates. While further validation by full-scale probabilistic risk assessments is necessary, we conclude that non-aviation users willing to accept average risk would obtain much better performance and availability from simple modifications to the existing SBAS and GBAS protection level calculations specified for aviation users.
Acknowledgments
We thank the FAA Satellite Navigation Program Office for its support of our research on WAAS and LAAS. However, the opinions expressed here are solely our own. We thank Jim Kelly and Tim Murphy for their explanations of the evolution of today’s SBAS and GBAS integrity requirements. We also thank the FAA Technical Center for its efforts in collecting and publishing WAAS error data over the last decade using its Performance Analysis Network (PAN).
Sam Pullen is a senior research engineer at Stanford University, where he is the director of the Local Area Augmentation System (LAAS) research effort. He has supported the FAA and others in developing GNSS system concepts, requirements, integrity algorithms, and performance models since obtaining his Ph.D. from Stanford in Aeronautics and Astronautics.
Todd Walter is a senior research engineer in the Department of Aeronautics and Astronautics at Stanford University. He received his Ph.D. from Stanford and is currently working on the Wide Area Augmentation System (WAAS), defining future architectures to provide aircraft guidance, and on assuring integrity on GPS III.
Per Enge is a professor of aeronautics and astronautics at Stanford University, where he is the Kleiner Perkins, Mayfield, Sequoia Capital Professor in the School of Engineering. He directs the GPS Research Laboratory and received his Ph.D. from the University of Illinois.
First Overload Interference/Desensitization to GPS Receivers, Systems, and Networks Report to FCC
The joint working group co-led by the U.S. GPS Industry Council and Lightsquared, investigating potential problems of LightSquared/GPS interference, delivered its first monthly report on March 15 as directed by the FCC. The report (PDF) lays out a schedule for receiver selection and testing and names 34 members, two working group co-chairs, and four information facilitators of a technical working group (TWG) supervising and analyzing the assessment of GNSS receivers operating under conditions of a dense national network of high-powered cell-phone transmitters. “TWG members represent a diverse group of interested parties including equipment and chipset manufacturers, aerospace/aviation companies, wireless providers, engineering firms, public safety, and various federal agencies. Additionally, several individuals have volunteered to be advisors to the TWG,” said the report.
The TWG held its first meeting on March 3 in Arlington, Virginia, and via a conference bridge for members around the globe who were unable to attend in person. In that and subsequent teleconferences, the TWG focused on the first seven items from the Work Plan:
Establish pertinent analytical and test methodologies and assumptions underlying the test regime: definition of harmful interference, relevant information regarding terrestrial broadband network, interference analysis assumptions, and evaluation of potential test methodologies.
Select categories of receivers and receivers to be tested.
Develop operational scenarios.
Establish methodology for analyzing test results.
Derive test conditions based on the established operational scenarios.
Write test plan and procedures.
Identify and engage appropriate test facilities.
LightSquared provided technical details to the TWG regarding the equipment planned for its terrestrial broadband deployment, including the channelization plan, output power, out-of-band emission (OOBE) characteristics, and emissions mask.
The GPS community is concerned that desensitization/overload due to strong signals outside of the GPS band may cause GPS receivers to operate in a non-linear mode with reduced gain (that is, gain compression) for the desired GPS signal. Other receiver impairments may also arise as a result of the nearby strong signals.
The TWG has agreed to move forward with a combination of laboratory-based and field-based testing programs. Field testing will be performed at outdoor test locations using transmitters, filters, and antennas similar to those that LightSquared plans to deploy in its commercial operations.
Other items of interest in the report:
Definition of Harmful interference at the GPS/GNSS/Augmentations/L-Band Receiver. “The TWG members have discussed a number of receiver parameters related to the definition of harmful interference. In the FCC Rules, harmful interference is defined as ‘interference which endangers the functioning of a radionavigation service or of other safety services or seriously degrades, obstructs, or repeatedly interrupts a radiocommunication service operating in accordance with [the ITU ] Radio Regulations.’
“Harmful interference affects different types of receivers in different ways. The key factors that pertain to the functioning of GPS receivers and/or whether service is degraded, obstructed, or interrupted are accuracy (position, velocity, time), availability (ability to perform a given function), coverage (within what space can a function be performed), integrity (what is the probability that the results are correct), and continuity (what is the probability that a given function can be completed). Metrics for harmful interference are developed from an understanding of the consequential relationship between negative impacts and receiver parameters, which include effective C/N0, PVT accuracy, time to first fix, loss of lock, cycle slips, etc. The signal conditions to be taken into account are defined in the GPS Standard Positioning Service (SPS) Performance Standard, 4th Edition, Interface Specifications (ISs), GPS policy, and both the present and planned future signal environments will be considered.Environmental and field conditions in which GPS receivers operate will also be considered.
“It should be possible to assess interference impact, up to that which includes harmful interference, using metrics in terms of receiver parameters that include measurable changes in effective C/N0 as well as position accuracy, time to first fix, loss of lock, cycle slips, etc. Related to this discussion is whether there is any margin that could be budgeted for terrestrial broadband operation, and if so, what that amount could be. When considering systems guaranteed for safety-of-life operations, there may be very little or no margin.
“There is general agreement within the TWG that the device testing protocols should include changes in effective C/N0 and degradation of other key performance measures so as not to exclude data that might be relevant for the post-testing analytical phase using operational scenarios.
Overload interference/desensitization at the GPS/GNSS/Augmentations/L-band Receiver. “Desensitization/overload due to strong signals outside of the GPS band may cause the GPS receiver to operate in a non-linear mode with reduced gain (i.e., gain compression) for the desired GPS signal; there may also be other receiver impairments caused by strong signals outside the GPS band. The TWG will consider these mechanisms further after testing is underway and sufficient samples are available to adequately assess such mechanisms.”
Evaluation of Potential Test Methodologies. “The TWG has agreed to move forward with a combination of laboratory-based and field-based testing programs. Laboratory tests are repeatable, allow for the creation of a fully controlled environment and the ability to test multiple scenarios and many devices in an efficient, repetitive manner. Field tests expose devices to a real-world environment where measurements can be performed at various distances and morphologies from terrestrial broadband network sites in order to gauge the effects of distance and physical environments on terrestrial broadband signal strength and potential interference. One advantage of field testing is that it captures a complete, live test environment comprehensively and helps develop keener testing or analysis insights that modeling cannot offer. The major disadvantage or concern is that field testing uses the present environment, not the environment that might exist at some future or past time. Interference testing analysis has to consider worse-case assumptions, and not only the current test reality.
“Laboratory testing will be performed either using conducted testing, where devices are connected directly to transmission sources via 50 ohm connectors, or through radiated testing in anechoic or other radiated emissions chambers. While conducted testing is the preferred laboratory methodology, anechoic chambers will be used where conducted testing is not practical, is not recommended by the manufacturer, or where connectorized devices cannot be made available within the established test timeline.
“Field testing will be performed at outdoor test locations that will utilize transmitters, filters, and antennas similar to those that will be deployed by Lig
htSquared in its commercial operations.”
The TWG identified seven categories of receivers that it considers representative of non-military GPS user equipment operating in the United States: aviation, cellular, general location/navigation, high precison, timing, space-based receivers, and networks.
Seven sub-teams are focusing on these receiver categories. The sub-teams are responsible for determining device selection and prioritization criteria, defining operational scenarios, listing testing conditions and test plan procedures, and recommending appropriate test facilities.
Save Our GPS Coalition Forms
Representatives from a variety of industries and companies have formed the Coalition to Save Our GPS to resolve what it terms a serious threat to the national positioning, navigation, and timing service: the FCC conditional waiver to Lightsquared allowing expansion of terrestrial use of the satellite spectrum immediately neighboring that of GPS, potentially causing severe interference to millions of GPS receivers.
“GPS is essential to Americans every day — it’s in our cars, the airplanes in which we fly and the ambulances, police cars, and fire trucks that help keep us safe. It’s also used in many industrial applications and even synchronizes our wireless, computer, and utility networks,” the group stated. “LightSquared’s plans to build up to 40,000 ground stations transmitting radio signals one billion times more powerful than GPS signals as received on earth could mean 40,000 ‘dead spots’ — each miles in diameter — disrupting the vitally important services GPS provides.”
The Coalition (www.SaveOurGPS.org) includes representatives from aviation, agriculture, transportation, construction, engineering, surveying, and GPS-based equipment manufacturers and service providers.
Initial members of the coalition are the Aeronautical Repair Stations Association, Air Transport Association, Aircraft Owners and Pilots Association, American Association of State Highway and Transportation Officials, American Rental Association, Associated Equipment Distributors, Association of Equipment Manufacturers, Case New Holland, Caterpillar Inc., Edison Electric Institute, Esri, Garmin, General Aviation Manufacturers Association, Deere & Company, National Association of Manufacturers, OmniSTAR, and Trimble. More members are expected to join in the near future.
The following is from a statement issued by the coalition:
“[In] The unusual waiver granted in January to LightSquared by the FCC . . . the usual FCC process of conducting extensive testing followed by approvals was not followed. Instead, the process was approve first, then test. Additional safeguards are needed, so the coalition recommends:
“The FCC must make clear, and the NTIA must ensure, that LightSquared’s license modification is contingent on the outcome of the mandated study. The study must be comprehensive, objective, and based on correct assumptions about existing GPS uses rather than theoretical possibilities.
“The FCC should make clear that LightSquared and their investors should not proceed to make any investment in operating facilities prior to a final FCC decision (or at least make it explicit that they do so at their own risk). While this is the FCC’s established policy, it failed to make this explicit in its order.
“Further, the FCC’s, and NTIA’s, finding that ‘harmful interference concerns have been resolved’ must mean ‘resolved to the satisfaction of preexisting GPS providers and users.’ Resolution of interference has to be the obligation of LightSquared, not the extensive GPS user community of millions of citizens. LightSquared must bear the costs of preventing interference of any kind resulting from operations on LightSquared’s frequencies.
“This is a matter of critical national interest. There must be a reasonable opportunity for public comment of at least 45 days on the report produced by the working group.”
WAAS Official Again
The Federal Aviation Administration (FAA) announced on March 18 that WAAS PRN 135 has resumed normal operations. “The WAAS team recently received the final report from Lockheed Martin on the failure of Galaxy 15,” reported FAA GNSS program manager Leo Eldredge. “After a review of that report, the team determined that the satellite was ready to be returned to operations.”
The FAA said that PRN 135 is currently located at ~120°W and enroute to its final destination of 133.1°W, but is now broadcasting operational corrections that can be used by both aviation and ground users, including those in Northwest Alaska.
In April 2010, satellite operator Intelsat reported it had lost contact with PRN 135 (named Galaxy 15) and it was drifting uncontrolled. At that time, the FAA reported that it would drift out of WAAS service within a few weeks. Instead, PRN 135 remained within a usable condition/location, although drifting east, until December 2010, when it ceased operating. On December 23, Intelsat reported that the power from the Galaxy 15 battery completely drained during its loss of Earth lock and the baseband equipment command unit reset, as it was designed to do. Shortly thereafter Galaxy 15 began accepting commands, and Intelsat engineers began receiving telemetry in the operations center.
Intelsat determined that static electricity charge caused the initial failure, and has uploaded new software to prevent the event from occurring again. There are now three operational WAAS GEO satellites:
◾ PRN 133 located at 98°W.
◾ PRN 135 located at 133.1°W (currently at ~120°W); will arrive at 133.1°W on or about April 4, 2011.
◾ PRN 138 located at 107.3°W.
EGNOS SOL Operational
The European Geostationary Navigation Overlay Service (EGNOS) was declared operational for safety-of-life (SOL) services on March 2. The service consists of GPS corrected signals intended for transport applications, particularly aviation, where lives could be endangered if the performance of the navigation system is degraded.
The SOL coverage area, expected performances, and conditions of use are described in the EGNOS Safety-Of-Life Service Definition Document (SDD, see env-gpsworld-integration.kinsta.cloud/egnosSOL). The two operational EGNOS satellites — Inmarsat-3-F2/AOR-E at 15.5 degrees west longitude using PRN code 120, and Artemis at 21.5 degrees east longitude using PRN code 124 — now transmit Message Type 2, indicating that the signals are available for safety-critical purposes.
Air-navigation service providers can now publish SBAS precision approach procedures, localizer performance with vertical guidance (LPV), based on EGNOS. On March 22, EGNOS operator European Satellite Services Provider published the first EGNOS LPV approaches for use at Pau Airport, near the Pyrénées in southern France.
EGNOS improves accuracy and provides integrity to the GPS signal over most of Europe and parts of North Africa. The system uses a monitoring network of 40 ground stations to provide the corrections with 99.9 percent availability over the core service region. Accuracy is measured by GPS user equivalent range error typically about 4.2 meters after EGNOS corrections for GPS signals from satellites at a 5-degree elevation, and 2.4 meters for satellite signals arriving from a 90-degree elevation. If reliability falls below a minimum level, EGNOS users are alerted within six seconds.
Russian SBAS Satellite Passes Transponder Tests
The Luch-5A geostationary communication satellite under construction has successfully completed a cycle of transponder tests. The satellite includes a transponder for the System for Differential Correction and Monitoring (SDCM), the Russian satellite-based augmentation system. SDCM will provide integrity monitoring of
GPS and GLONASS satellites and differential corrections and analyses of GLONASS performance: real-time differential corrections with horizontal accuracy of 1–1.5 meters, vertical of 2–3 meters.
There’s something I’ve been wanting to write about since the ION-GNSS conference a few weeks ago. However, a nasty cold, a 10-day trip to Europe (INTERGEO conference), and some jet lag have kept me from it until now.
Here goes.
First of all, most of the presentations from the CGSIC meeting are available on the USCG Navigation Center website. You can view them by clicking here. There’s some very good reading and most of it is pretty light-weight and in PDF format.
One of the presentations at the CGSIC (Civil GPS Service Interface Committee) meeting during the ION-GNSS conference was “Integrating NDGPS and SBAS —
An Optimal Real-time GPS Mapping Solution,” presented by Jean-Yves Lauture of Geneq, Inc.
I’m publishing two of the slides from his presentation in order to:
Show the accuracy potential of WAAS and NDGPS given a high performance L1 receiver.
Discuss the statistical names/values used to express GPS accuracy.
First of all, each of the slides below are at the same scale. Each ellipse is 20 cm with the outside limit (radius) being one meter.
I’ve known for quite sometime that SBAS (WAAS in this case) is capable of sub-meter precision with a single-frequency GPS receiver. These results are a bit better than what I’ve seen personally, and keep in mind it’s a limited data set of 1,800 continuous epochs, but impressive none the less. Also, keep in mind that the WAAS Performance Analysis Report published quarterly by the FAA’s National Satellite Test Bed shows the 95% horizontal accuracy value for Denver, Colorado, (near where this data was collected) being .547 meters for the quarter ending June 30, 2010 (7,856,354 samples collected over three months).
30 minutes of WAAS-corrected data (each ellipse represents 20cm)
The results I didn’t expect were the slide below, which shows NDGPS-corrected results using the same receiver/antenna. Keep in mind this is a GPS L1 receiver using phase-smoothed pseudorange measurements, not a GPS L1/L2 receiver using a carrier-phase float solution. If you look closely, you’ll see it states the baseline distance is 200 km. Granted, this is a limited data set, and I’ll be interested in seeing further results. If this was a dataset presented by a manufacturer or other party with some sort of interest, I wouldn’t publish it, but this is data collected by an objective entity (a credible U.S. government agency) so that earns, in my mind, a level of credibility.
The results are pretty impressive. All data points fall within ~20 cm.
30 minutes of NDGPS-corrected data (each ellipse represents 20cm)
Keep in mind that this data was collected recently, and we are currently in a period of low ionospheric activity. In other words, data was collected under near-ideal conditions. At the end of the day, my point is that GPS L1 accuracy using SBAS and NDGPS has gotten pretty darned good.
Accuracy Statistics
The second reason I’m publishing the slides is to discuss accuracy statistics.
Look at the small box inside each slide showing 99%, 95%, 68%, and 50% accuracies.
If you look at the data points, it might not be immediately apparent how those values were arrived at. For example, how could a group of data points all within ~20 cm have a 95% confidence of 37 cm?
To explain this, there was a good article published in GPS World in 2007 titled “GNSS Accuracy: Lies, Damn Lies, and Statistics” by Frank van Diggelen. It does a good job explaining statistical expressions (RMS, 2DRMS, etc.).
Keep in mind that most manufacturers express horizontal GPS accuracy specifications based on 68% confidence. When the specification sheet states “sub-meter” HRMS (horizontal RMS) precision, that means 68% of the time; the horizontal accuracy will be less than a meter. In reality, that “sub-meter” receiver won’t consistently deliver sub-meter precision. If you convert the 68% HRMS value and express it with 95% confidence (2D HRMS), the actual horizontal precision for that same receiver will be well over one meter. That’s the precision you can expect from the receiver, not the 68% confidence value.
After reviewing current performance of WAAS, EGNOS, and MSAS, the authors present expected future performance, including the benefits of GPS L5. They evaluate the impact of the Indian GAGAN and Russian SDCM systems on global coverage and examine southward expansions for the original three SBASs. Finally, a look at the impact of a second constellation of navigation satellites and the performance for a user taking advantage of two core constellations.
By Todd Walter, Juan Blanch, and Per Enge, Stanford University
The Wide Area Augmentation System (WAAS) monitors GPS and provides both differential corrections to improve accuracy and associated confidence bounds to assure integrity. The first satellite-based augmentation system (SBAS), it was commissioned for service in 2003. Japan’s MTSAT-based Satellite Augmentation System (MSAS) was commissioned in 2007, and the European Geostationary Navigation Overlay Service (EGNOS) was declared operational in 2009, with safety-of-life service commissioning expected in mid-2010. Two other SBASs are in the developmental stage: India’s GPS Aided Geo Augmented Navigation (GAGAN) and Russia’s System for Differential Corrections and Monitoring (SDCM) have fielded equipment and plan to become operational in the next few years.
Coming improvements will expand SBAS coverage areas and strengthen their performance. In the near term, these include more monitoring stations and algorithmic enhancements, with incorporation of a second civil signal in a protected aeronautical band and new GNSS constellations in the long term.
An SBAS utilizes a network of precisely surveyed reference receivers, located throughout its coverage region. The information gathered from these reference stations monitors the GNSS satellites and their propagation environment in real time. Availability of SBAS service is a function of two quantities: the arrangement of the pseudorange measurements used to determine the user’s position, referred to as geometry; and the quality of each individual measurement, referred to as the confidence bound. Although very small confidence bounds can make up for poor geometries, and strong geometries can overcome large confidence bounds, both values are generally required to be good to obtain high availability.
Geometry is determined purely by the locations of the ranging satellites relative to the user. Currently the basic geometry is provided by the GPS constellation. Historically it has exceeded commitments, and there are currently 29 healthy satellites in orbit when only 21 are nominally guaranteed. However, as satellites are taken off-line in critical orbital slots, the quality of the geometry can degrade significantly. There could be short duration losses of service daily at some locations. Since the goal is to provide service more than 99.9 percent of the time, these outages can have a dramatic impact. WAAS currently mitigates this concern by adding geostationary satellites with a ranging function virtually identical to the GPS satellites. These satellites are always in view and improve the overall geometry, although they do not eliminate the problem completely.
The confidence bounds relate to the expected error sources on the range measurements. Currently three error sources are corrected via broadcast to the user: satellite clock error, satellite ephemeris error, and delay error due to propagation through the ionosphere. These error sources are described by two confidence bound terms: the user differential range error (UDRE) for the satellite errors, and the grid ionospheric vertical error (GIVE) for the ionospheric errors.
For single-frequency SBAS, this last error source is the most significant. Users may sample the ionosphere anywhere in the service volume, but the SBAS only has measurements from its reference station locations. Thus, there is always the possibility of undetected ionospheric disturbances. This leads to larger confidence bounding terms and lower availability.
The combination of geometry and confidence bounds yields the protection levels (PL). PLs are the real-time confidence bound on the user’s position error. To match aviation requirements these are broken into a vertical protection level (VPL) and a horizontal protection level (HPL). Each SBAS guarantees that the user’s actual position error will be smaller than these values 99.99999 percent of the time. The PLs are calculated in real-time using stored and broadcast information. They must be compared to the maximum allowed value for a desired operation. The upper bounds are called alert limits (AL) and they are fixed numbers whose values depend on the operation.
In this article we are interested in the localizer performance with vertical guidance (LPV)-200 approach with a VAL of 35 meters and HAL of 40 meters. Currently, LPV aviation approaches can only be accomplished with a WAAS GPS receiver. Performance of an LPV approach allows minimums as low as 200 feet above ground level before a missed approach must be executed. As of January 2010, there were 1,930 published WAAS LPVs, with plans to add 300 per year in the United States.
Because GPS and SBAS generally perform better at horizontal positioning than vertical, the requirement that the VPL be below the VAL is nearly always the limiting constraint for these operations.
Methodology
To determine the global availability and the effect of potential improvements, we used our Matlab Algorithm Availability Simulation Tool (MAAST). This tool uses almanac data to calculate the position of the satellites for each specified epoch. The almanac chosen for this study corresponds to the GPS almanac broadcast on April 8, 2009, when there were 30 healthy satellites. However, PRNs 25 and 32 were removed to simulate a condition with 28 healthy satellites. MAAST also implements the WAAS integrity algorithms to calculate the corresponding UDRE and GIVE values. Finally, it uses these values to implement the airborne algorithms specified in the minimum operational performance standards (MOPS) for SBAS. The MOPS specifies user algorithms for determining the protection levels. For these simulations, the VPL and HPL are calculated about every 5 minutes and every two and a half degrees across the globe.
MAAST does a good job of predicting WAAS behavior. It is less accurate when predicting other systems’ performance. EGNOS has developed its own monitoring receivers and integrity algorithms and has different criteria for assigning a satellite a particular UDRE value and assigning each ionospheric grid point’s (IGP’s) GIVE value. Nevertheless, both systems are designed to meet ICAO requirements for integrity, and their performance should be somewhat similar. In observing EGNOS coverage plots and comparing them to MAAST predictions, we do see differences. However, the size of the coverage region and approximate boundaries are reasonably close and provide an idea of performance if not an exact map.
The MSAS algorithms are based upon the same algorithms used in earlier versions of WAAS. Therefore, MAAST should be slightly more accurate in modeling its performance. GAGAN uses the same prime contractor as WAAS and therefore similar algorithms may be expected. Less is known about the intended SDCM algorithms and therefore the modeling of this system faces the largest uncertainty. Again, the MAAST predictions should be viewed as indicative rather than precise. Individual availability maps will not be completely correct, but relative performance improvements should be properly indicated.
Current Systems Status
Currently WAAS is in its full LPV-200 performance (FLP) phase. It consists of 20 WAAS reference stations (WRS) in the conterminous United States (CONUS), in addition to seven in Alaska, one in Hawaii, one in Puerto Rico, four in Canada, and five in Mexico for a total of 38. The station locations are shown as blue circles in Figure 1. There are three WAAS master stations (WMS) and two geostationary satellites (GEOs). The GEOs are the Intelsat Galaxy XV satellite
at 1338 W and the Telesat ANIK F1R satellite at 1078 W.
FIGURE 1. Existing SBAS reference networks, consisting of 38 reference stations for WAAS, 34 for EGNOS, and 8 for MSAS.FIGURE 2. Simulation results from MAAST for availability of LPV-200 provided by current systems.
As can be seen in Figure 2, availability of LPV-200 service is very high for most of North America. In general, this performance meets the goals for the system. However, in some regions performance is lower than the 99 percent minimum target. The West Coast, Alaska, and Southern Mexico all suffer from reduced availability.
MSAS is in its initial operating phase. It consists of six ground monitoring stations (GMSs) on the Japanese Islands, one in Australia, and one in Hawaii (magenta triangles in Figure 1). There are two master control stations (MCSs) and two Multifunction Transport Satellite (MTSAT) geostationary satellites at 1408 E and 1458 E.
Because of the limited network size, the GEO UDREs for MSAS are set to 50 meters and therefore do not benefit vertical guidance. Further, the limited ionospheric observations offer little availability of LPV-200 service as can be seen in Figure 2. As a result, vertically guided operations have not yet been authorized based upon MSAS. The Japanese Civil Aviation Bureau (JCAB) has studied performance improvements that could allow it to provide LPV-200 operations. Until then, MSAS provides only lateral navigation.
EGNOS is also in its initial operations phase. It consists of 28 ranging and integrity monitoring stations (RIMS) in Europe, one in Turkey, three in Africa, one in North America, and one in South America (green squares in Figure 1). There are four master control centers (MCCs) and two GEOs, the INMARSAT Atlantic Ocean Region-East (AOR-E) satellite at 15.58 W and the ARTEMIS satellite at 21.58 E.
For a variety of reasons, EGNOS has chosen to implement its GEO satellites without a ranging capability. Thus, for our simulations we have set them as data-links only and do not model a ranging capability. EGNOS also currently implements Message Type 27 (MT-27) rather than Message Type 28 (MT-28) as do WAAS and MSAS. MT-27 restricts the use of low UDRE values to a box centered on the European region. Its borders can be discerned in Figure 2. Currently it has little impact on LPV-200 service, but if EGNOS is to expand its coverage, it may become a limiting factor. Availability of LPV-200 service is very high for most of Europe. However, there is a desire to expand coverage to more reliably cover Iceland, Scandinavia, Eastern Europe, and the Mediterranean and South Atlantic regions.
Near-Term Improvements
EGNOS is fielding additional reference stations in the Canary Islands, Northern Africa, and the Middle East. In the longer term, MT-28 is being considered as a replacement for MT-27. In our modeling we added seven new RIMS, shown in Figure 3, and implemented MT-28. We also improved the ionospheric mask by including additional IGPs. We did not update GEO locations nor did we model ranging capability that could further enhance performance. By comparing
FIGURE 3. Improved SBAS networks. The newly added reference stations are marked by yellow filled squares for EGNOS and yellow filled triangles for MSAS.
Figure 4 to Figure 2 improvements can be seen, in particular expanded LPV-200 operation to the south.
FIGURE 4. Improved single frequency SBAS coverage for the original three SBAS.
The future of MSAS improvements is less certain, with no firm commitments for major service enhancements. We have chosen to model fairly aggressive enhancements based upon studies made by the Electronic Navigation Research Institute in Japan. We have added 10 new reference stations in Japan and made the ionospheric threat model less conservative, in line with current WAAS algorithms. Together, these improvements offer good vertical guidance coverage over Japan.
These improvements extend coverage in the vicinity of the reference station networks, but are unable to push availability much beyond. This is primarily due to the limitations of the ionospheric corrections. Because strong gradients can exist outside of the viewing area of the network, tight confidences cannot be provided to those regions.
SBASs model the ionosphere as a thin 2-dimensional shell 350 kilometers above Earth. This works well for quiet mid-latitude and polar ionosphere. However, equatorial ionosphere often has significant vertical structure that is not well replicated by the SBAS message. The resulting confidence bounds are then too large to reliably provide LPV-200 capability. No certified algorithm capable of bounding the equatorial ionosphere is known to the authors. Instead, it is recommended that SBASs in equatorial areas wait for the forthcoming L5 signal to provide vertical guidance in their regions.
GPS L5
The next GPS satellite to be launched will contain a new civil signal, L5, centered at 1176.45 MHz and in a protected aviation band. As such, it will be approved for use on aircraft. When the L5 signal is used in combination with L1, the ionospheric delay for each line-of-sight can be directly estimated. This will dramatically lower the uncertainty of the pseudorange measurement. Thus, if the SBAS is upgraded to provide corrections appropriate for an L1/L5 user and the user similarly upgrades his or her avionics, SBAS service can be dramatically improved.
Another important advantage of the second civil frequency is its relative immunity to ionospheric storms. Because the users are now directly eliminating the amount of delay they actually experience, they are no longer affected by shortcomings in the MOPS ionospheric model. The weaker effect of scintillation may have some impact; however, we do not expect to lose vertical guidance altogether. Furthermore, the availability of two civil frequencies offers protection against unintentional interference. If either L1 or L5 is jammed, the user still has access to guidance on the available frequency.
At the moment there is no MOPS for an L1/L5 user, so any ground or user algorithms will have to be speculative. We propose basing future L1/L5 algorithms on the existing L1-only algorithms. Instead of using L1-only pseudorange measurements, the user forms the ionosphere-free combination. For the confidence term representing the total pseudorange error on a line-of-sight, the ionospheric correction terms and airborne multipath terms are replaced with a single value representing airborne noise and multipath for the ionosphere-free combination.
For a single frequency user, each line-of-sight has four confidence terms that are summed together to obtain the total confidence. These terms correspond to: the satellite clock and ephemeris corrections (σflt), the ionospheric correction (σUIRE), the airborne code noise and multipath (σair), and the troposphere (σtrop). The total one-sigma confidence bound for a particular line-of-sight is the root sum square (RSS) of these four terms:
(1)
When a user has access to two civil frequencies, they can remove the ionospheric effects by forming the iono-free combination of the two pseudoranges:
(2)
where f1 and f5 are the L1 and L5 frequencies (1575.42 MHz and 1176.45 MHz) respectively. If σ1 and σ5 are comparable then the iono-free combination has roughly three times as much noise as either single frequency term, but is substantially smaller than σUIRE . Furthermore, satellites do not need a grid correction to be used, thus satellites farther from the network and IGP mask can be incorporated into the position solution. The dual-frequency confidence bound for a single satellite is then given by
(3)
where σair is used in place of σ1 and σ5 in (2).
For the VPL we propose adding nominal bias terms to handle observed signal biases and non-Gaussian behavior of the underlying error terms. By including these terms it is possible to reduce the net impact of these biases on the user. Further, we propose tailoring the VPL equation to the most significant remaining threat to the user: single satellite fault modes. The L1-only VPL equation is appropriate for threats that affect many signals simultaneously as may happen with the ionosphere or troposphere. However, with the user directly eliminating ionospheric effects, the most significant threats come from satellite fault modes. As these faults are rare, they are unlikely to affect more than one ranging measurement at a time. Therefore, a VPL can be constructed to explicitly account for such a threat. We recommend that the dual frequency VPL take the following form:
(4)
where KHMI and σ5 is the Gaussian tail factor corresponding to the probability of Hazardously Misleading Information, s3,i is the projection of the pseudorange error onto the vertical position estimate, sff is the fault free overbounding sigma, biasnomis the nominal bias bound, Kfault is the Gaussian tail factor accounting for the probability of fault, and biasfault is a bound on the magnitude of all satellite faults. The H0 condition corresponds to the most likely condition of no faults present. The H1 condition corresponds to the unlikely event of a fault on the dominant satellite. The final VPL is the maximum across both conditions.
Because the faulted bias term covers the satellite faults the fault-free sigma term, σff, can be much smaller than the current total value (1), or the dual frequency version (3). Further, since the probability of fault is small, Kfault can be much smaller than KHMI . The net result is that the proposed VPL is smaller than the existing VPL for the same conditions. To model L1/L5 availability we chose the following parameters:
KHMI = 5.33
Kfault = 2.33
σ 2ff = (σflt / 3 ) 2 + σ 2iono_free + σ 2trop
biasnom = 0.5 m
biasfault = 5.333 x σflt
Other values follow the single frequency MOPS specifications as normally implemented by MAAST.
Given these parameters, the H1 hypothesis nearly always dominates the VPL calculation. We have used a nominal weighting scheme to optimize for accuracy. It is possible to deweight the dominant satellite to improve availability. We will be looking at practical methods for determining more optimal weighting for the VPL given in (4). However, there is a concern that such optimizations could harm accuracy. The potential benefits vs. risks will be studied.
The improvement in performance for a dual-frequency user can be seen in Figure 5. The coverage is significantly expanded. Now each region is robustly covered with large margins surrounding their intended service regions. However, coverage is still limited to the areas around these first three SBASs.
FIGURE 5. Potential dual frequency coverage of the first three SBASs including network improvements.
GAGAN and SDCM
Two additional SBASs are currently under development that will extend coverage to more regions. India is developing GAGAN. Currently it has eight Indian reference stations (INRES) all in India (blue diamonds in Figure 6). There is one Indian master control center (INMCC), and plans to use the GSAT-4 as its initial GEO. The GSAT-4 is planned for launch in 2010 and will be located near 82° E. The geomagnetic equator passes through India and it therefore faces the full impact of equatorial ionosphere. The advent of L5 will allow GAGAN to obtain high LPV-200 availability that is unlikely to be achievable for single-frequency users.
FIGURE 6. The networks of five SBAS systems are shown. In addition to the reference stations from Figure 3, the 8 Indian stations are shown as blue diamonds and the 19 Russian stations are shown as red stars.
Russia is developing SDCM. It now has nine operational measuring points (MPs) and has plans for at least 10 more locations, all in Russia (red stars in Figure 6). There are also plans to use three GEOs: Luch-5a planned for launch in 2010 and to be located near 16° W, Luch-5b planned for launch in 2011 and to be located near 95° E, and Luch-4 planned for launch in 2013 and to be located near 167° E.
Figure 7 shows the combined dual-frequency coverage of all five systems, WAAS, EGNOS, MSAS, GAGAN, and SDCM.
FIGURE 7. The combined dual frequency availability of the five SBASs is shown.
The vast majority of land masses in the northern hemisphere are now well covered by at least one of the SBASs. Figures 6 and 7 clearly highlight that the majority of development has occurred in the northern hemisphere. In fact, only two reference stations have been placed below the Equator.
Southern Hemisphere
If SBAS is to provide a global solution, its coverage must extend into the southern hemisphere. There have been many discussions with representatives of countries in the southern hemisphere. Further, the United States has had testbed receivers in South America for nearly 15 years. Europe has fielded receivers in Africa. Australia investigated its own variant of SBAS called the Ground-based Regional Augmentation System (GRAS). However, we are not aware of concrete plans for development in this hemisphere.
We anticipate that discussions will eventually evolve into firm plans and that either independent SBASs will be developed in these regions or existing SBASs will expand their networks southward. We have chosen to assume that WAAS, EGNOS, and MSAS will expand their networks to extend LPV-200 coverage to the southern portion of their GEO footprints. This is but one of many possible scenarios. The pr
oposed expansion shown in Figure 8 is not based on any plans, but is based on the notion that civil aviation authorities will want to obtain global coverage. The assumed new southern reference stations are shown as yellow-filled circles for WAAS in South America, yellow-filled squares for EGNOS in southern Africa, and yellow-filled triangles for MSAS in and around Australia. Advantages of dual frequency allow us to have much less dense networks for the expansions, in addition to allowing LPV-200 capability to be obtained in equatorial areas.
FIGURE 8. The networks of the five SBAS systems including hypothetical expansions into the southern hemisphere
Figure 9 shows the combined dual-frequency coverage for these SBASs with the expanded network. Now nearly all land masses have good LPV-200 coverage. Note that we have not attempted to optimize these networks to assure coverage to all land masses, not have we tried to find the minimum number of stations that offer this capability.
FIGURE 9. The combined dual frequency availability of the SBASs with the southern hemisphere stations is shown.
Added Core Constellations
Galileo is envisioned as compatible with GPS in that each satellite provides ranging using signals covering the L1 and L5 frequencies with similar modulations. Although the final specifications are not yet set, it is envisioned that Galileo satellites will provide a service that is fully interoperable with the GPS civil signals. Thus, we can approximately model Galileo satellites as being equivalent to GPS satellites in different orbits. In parallel, China is developing the COMPASS system whose signals are also planned to be compatible with GPS.
The Russian GLONASS system has been operational for many years. However, its current signal structure makes it less suited for incorporation into avionics. There are modernization plans to broadcast L1 signals that are more in alignment with the other constellations. Thus it, too, may one day be incorporated into SBAS. We believe that SBASs will someday broadcast satellite clock and ephemeris corrections for GPS and one or more other core constellations. These corrections will remove any difference in the reference times or coordinate frames between the two systems, allowing the corrected signals to be considered fully interchangeable.
Adding 24 or more extra ranging sources will have tremendous benefit for all civil GNSS users. The user’s geometry would be very robust to the loss of one or two satellites. Adding one or more core constellations has the potential to significantly improve SBAS coverage. We chose to model the addition of one constellation, by combining the almanac we used for GPS with one that had been proposed for Galileo. For these scenarios, MAAST is modeling 55 medium earth orbiting navigation satellites in addition to the GEOS used by each SBAS. Because the orbital repeat period is approximately 10 sidereal days for Galileo, the simulated time step and total run time were each increased by a factor of ten.
Figure 10 shows the improved coverage when the reference stations shown in Figure 6 are used. The additional satellites fill in many potential coverage gaps and now, compared to Figure 7, the SBASs all have even more reliable coverage well beyond their reference networks. Indeed, the Northern Hemisphere is now essentially fully covered. Figure 11 shows the results when the expanded networks of Figure 8 are incorporated. Compared to Figure 9, the southern hemisphere is much more reliably covered. The remaining gaps could easily be filled in with just a few more reference stations if full global coverage were desired.
FIGURE 10. The combined dual-frequency, LPV-200 coverage of the five SBAS systems with both GPS and Galileo.FIGURE 11. Combined dual-frequency LPV-200 coverage, SBASs with GPS and Galileo and the southern hemisphere stations.
Conclusions
For single-frequency SBAS the coverage is limited to areas very close to the monitoring station network. However, each region can obtain very good LPV-200 coverage within their desired service area. The addition of GPS L5 makes vertical guidance largely immune to ionospheric disturbances, and permits SBAS coverage to extend into equatorial areas. Independence from the ionospheric grid also allows service to extend farther away from the core network regions. When new Indian and Russian systems are commissioned, a very large fraction of the northern hemisphere will have LPV-200 coverage.
With dual frequency, LPV-200 coverage can be established with comparatively sparse networks in South America, Africa, and around Australia. Additional dual-frequency core constellations such as Galileo, Compass, or GLONASS could greatly expand coverage to well outside the original reference network regions. As GNSS capability is improved and expanded, we anticipate that SBAS coverage may one day provide nearly global LPV-200 or better service capability.
Acknowledgments
The authors acknowledge support of the FAA Satellite Product Office. However, the opinions and potential future scenarios reflect those of the authors and are not necessarily representative of the FAA.
Todd Walter is a senior research engineer at Stanford University. He has been active in the development of the Wide Area Augmentation System and related systems around the globe. His focus is on the provision of certified integrity for aviation applications.
Juan Blanch is a research associate at Stanford University, where he works on integrity algorithms for GNSS. He holds a Ph.D. in aeronautics and astronautics from Stanford.
Per Enge is professor of aeronautics and astronautics at Stanford, where he directs the Stanford GPS Research Laboratory. He has a Ph.D. from the University of Illinois.
Yowza!!, an application designed for the latest GPS-enabled iPhone 3G and 3GS models and iPod Touch, brings relevant coupon offers to customers based on their location.
“Any time you insert a concept such as location into a marketing program, you end up with a far more compelling value proposition,” states Mike Wehrs, president of the Mobile Marketing Association.
Sales and discount offers via Yowza!! can be updated in real-time and targeted by region or store location. “The phone will deliver a list of stores within one mile that have offers on Yowza!!,” said August Trometer, co-founder of the recent startup. Users show the barcode and digital mobile coupon on their handset at checkout to redeem the discount on their purchase.
“We work directly with merchants; they provide us with their latitude and longitude, we get the GPS coordinates, do a database search with a proprietary algorithm,” said Trometer. “The phone constantly goes back and forth between our app, touching data from our database. When the person touches their location, it touches a new set of data in the database. The phone will work with them to keep delivering the closest store. There’s a lot of work on the database end of things.”
One drawback of the app is that it has to be turned on to work — it does not sit in the background, waiting to be activated by incoming offers. “Users have to give the application access to their GPS coordinates,” explained Trometer. “But the power of the device and all the applications it brings make it silly to turn off the location capability.”
Retailers that have signed with Yowza!! include Sears, McDonald’s, The Container Store, and more. Unlike traditional forms of couponing such as newspaper ads, Yowza!! offers can be updated in real time and targeted by region or store location.
Trometer expects to announce Yowza!! capability through other GPS-equipped phones: Blackberry Storm, Google’s Android-based phone, and the Palm Pre. “All three makers allow developer access to the GPS and this is very important, it’s crucial, obviously. They also have a high-res screen, which is a requirement for our scannable barcode that the user shows to the merchant.”
Referring to GPS handsets that lack a high-res screen, he claims “The other phone manufacturers really have an uphill battle right now.”
Whose GPS? The source of the GPS chip within Apple’s iPhone remains a mystery. “Even people who have done teardowns of the devices, the chips are completely blank,” says Trometer.
“There are so many possibilities, we’re just scratching the surface right now with what can be done,” Trometer said. “The mind reels with the things that can be done with that.”
>> SURVEY & CONSTRUCTION
Hemisphere, Juniper Jointly Offer DGPS Receiver for Demanding Environments
Juniper Systems and Hemisphere GPS offer the XF101 DGPS receiver for the Archer Field PC, designed to deliver sub-meter DGPS to location-based applications in demanding environments.
According to the companies, the Hemisphere GPS XF101 DGPS receiver provides: Crescent GPS technology for sub-meter accuracy; COAST technology to maintain accuracy during temporary loss of differential signal; optional external antenna for centimeter-level accuracy; low power consumption; modular connection for rapid field use; real-time or post-processed DGPS data collection; and multipath minimization.
The XF101 with the Archer is priced at less than $2,500. It fully supports mobile GIS applications such as ESRI ArcPad and OnPoz GNSS Driver.
>> AVIONICS
NovAtel Receiver for Next-Gen WAAS
NovAtel announced receipt of a contract from the U.S. Federal Aviation Administration (FAA) to develop the next generation Wide Area Augmentation System (WAAS) reference receiver, the GIII. Total contract value can go up to $9.7 million.
NovAtel has worked with the FAA WAAS program since 1995, providing and supporting two previous generations of reference receivers for the WAAS ground network. The technology refresh will add support for new L1C, L2C, and L5 signal capabilities, on a qualified RTCA DO-178B software and DO-254 hardware platform. The WAAS GIII receiver program is scheduled to be completed over the next three years, and will include growth provision for further signal capability such as Galileo. As many as 14 receivers will be produced in the GIII development and qualification program.
>> FLEET TRACKING
AT&T, Trimble Fleet Management
AT&T has broadened its fleet and mobile asset management portfolio with the latest version of Trimble’s GeoManager solution, which helps reduce fuel and maintenance costs by enabling operators to manage their vehicle assets more efficiently.
Trimble GeoManager enables transportation and field-service fleet operators to track their mobile workers and assets through software and GPS modems running on AT&T’s wireless network. GeoManager integrates GPS, wireless data communications, and a browser interface to help manage mobile workers, the mobile worker’s work, and the mobile worker’s assets.
AT&T and Trimble have jointly offered fleet-tracking solutions for several years. The GeoManager update features improved map and status, new landmark uploads, WLAN usage, schedule report enhancements, driver logs, and organizational hierarchy modifications.
>> TIMING
Timing Vulnerability Concern Grows
Industrial and enterprise users in telecommunications and utilities privately express concern over revelations from the April Government Accounting Office (GAO) report, “Global Positioning System: Significant Challenges in Sustaining and Upgrading Widely Used Capabilities.” The GPS signal is used for synchronizing almost all global computer networks belonging to the military, utilities, banks, telecomms, television companies, and many more.
Backup? What Backup? These same companies point to a continued lack of commitment on the part of the U.S. government to stable and reliable backup for GPS. As long ago as 2007, in comments before the Department of Transportation, wireless carrier Sprint Nextel stated: “Sprint Nextel Corporation respectfully requests that the U.S. government continue to operate and invest in the LORAN-C and eLORAN systems. Should the DOT and DHS decide to decommission the LORAN-C system, Sprint Nextel recommends that the agencies delay doing so until the eLORAN system is fully operational. Sprint Nextel and other communications providers use the frequency signals of the Global Positioning System, LORAN, and atomic clocks for multiple levels of redundancy and diversity in their networks. Therefore, Sprint Nextel urges the DOT and DHS to carefully weigh decisions which might impact LORAN’s availability to the nation’s voice and data communications networks.
“The loss of a primary reference source (PRS) can negatively impact a telecommunications network, and those impacts can vary from minor short-term noise impairments to long-term network-wide outages. Both traditional wireline services and newer wireless services require a precise frequency reference for basic service delivery . . . . The continental U.S. portion of the Sprint Nextel network requires a PRS at thousands of switch sites, interconnection sites
, and cell tower sites to ensure reliable service delivery.”
Deadlock on Capitol Hill. Competing resolutions to either discontinue or adequately fund LORAN and eLORAN continue fencing in Congressional subcommittees in both chambers. Nothing has changed since Sprint commented two years ago — aside from a potential rise in the susceptibility of GPS to jamming, unintentional interference, and decreased availability.
GAO REPORT, FIGURE 5. Probability of maintaining constellation of at least 18, 21, and 24 GPS satellites based on reliability data as of March 2009 and a two-year GPS III launch delay.
>> TIMING
Telecom Clock from EndRun
EndRun Technologies announced a Telecom Clock Option for its Meridian Precision GPS Timebase, which provides accurate and stable GPS-synchronized outputs for military communications, aerospace, broadcast, engineering and calibration laboratories, telecommunications, and more.
The option was designed as a plug-and-play module that can supply any combination of E1, T1, J1 and/or composite clock outputs. An alarm output is also available and single-satellite mode (SSM) is supported. The Telecom Clock Option can be installed in EndRun’s GPS or CDMA-based Meridian and Tycho product lines.
On December 6, 2006, the sun emitted a burst of radio energy that impacted the performance of GPS receivers all over the sunlit side of the Earth. That the sun produces radio emissions is not surprising. What is surprising is that on this day they were extremely powerful. The sun continuously emits energy across a broad region of the radio spectrum. The flux density of these emissions is normally fairly low and contributes imperceptibly to the background radio noise collected by GPS receiver antennas.
However, when a solar flare occurs, it is often accompanied by very powerful bursts of radio energy. Although they are more numerous near the peak of the solar sunspot cycle when the sun is more disturbed, solar flares and their associated strong radio bursts can occur at anytime – including near the current sunspot minimum. Still, the December 6 solar radio burst came as a surprise. It was one of the largest on record and had an impact on all GPS receivers on the sunlit side of the Earth, including most of North America, South America, and the Pacific Ocean. The added noise significantly reduced carrier-to-noise-densities (C/N0 – a measure of the strength of received signals) at both the L1 and L2 frequencies by as much as 15 dB-Hz. This resulted in receivers losing lock on some satellites for many minutes, particularly those at low elevation angles with low C/N0 values before the burst’s arrival. Those receivers closer to the sub-solar point were typically affected more than those further away as more or the burst energy was picked up by the receiver antennas.
Nevertheless, it appears that a lot of single-frequency receivers continued to provide navigation solutions with as few as four satellites — and even three in 2D mode — and the noise burst went unnoticed by most users of such receivers. However, many dual-frequency receivers used for high-accuracy applications including those at reference stations suffered significant signal losses, particularly at the L2 frequency. As well, military receivers in some sectors lost the ability to navigate. A “widespread loss of GPS” in the Four Corners region of New Mexico and Colorado was reported by military authorities. Several aircraft reported losing lock on GPS signals with the number of tracked satellites dropping from 7-9 to 1 or even none!
Alessandro Cerruti, a graduate student at Cornell University, is among a group of scientists and engineers studying the effects of this and other solar radio bursts on the operation of GPS receivers. He has examined the data provided by the receivers in the International GNSS Service (IGS) network on the sunlit side of the Earth. The number of stations providing data at both frequencies on at least four satellites dropped from more than 120 to below 60 during the burst. The timing of the drop-outs coincides with the power of the burst which is shown in the lower panel.
The burst power was recorded at the Owens Valley Solar Array (OVSA) in California’s high desert. Operated by the New Jersey Institute of Technology’s Center for Solar-Terrestrial Research, OVSA records solar radio emissions at over a range of frequencies and polarizations including right-hand circular polarization (RHCP) at 1.6 GHz, very near the GPS L1 frequency. As the plot shows, noise power exceed one million solar flux units at the peaks of the burst, making this burst one of the largest on record.
Alessandro Cerruti has also looked at data from the Wide Area Augmentation System (WAAS) which is very robust and although WAAS continued to operate throughout the period of the burst, signals at the WAAS reference stations suffered significant degradations as elsewhere. The C/N0 values for PRN 4 as recorded at the Houston reference station on both the L1 and L2 frequency for a quiet day and on December 6. The drop in C/N0 values during the burst is very dramatic.
Mitigation. What can be done, if anything, to mitigate the effects of solar radio bursts? As the bursts are broadband noise, it is difficult for a receiver to discriminate them from GPS signals. Some antenna designs, such as choke rings, attenuate signals arriving at low elevation angles, so if the sun is low in the sky at the time of a burst, receivers with such antennas will be less impacted than those with conventional antenna designs. And as a receiver loses track primarily on satellites at low elevation angles, having more satellites at higher elevation angles will also help. So receivers operating with a mixed constellation of GPS and GLONASS or GPS and Galileo satellites should be better able to weather a solar radio burst than those operating with GPS alone. Similarly, a larger GPS constellation by itself would help.
Modernization. Stronger transmitted signals from future GPS satellites might allow receivers to continue tracking even low-angle satellites during a large burst. Newer signal formats, which could be tracked at lower C/N0 values, would also help receivers to contend with the sun’s outbursts. Even current receiver technology developed for anti-jamming protection and for indoor GPS use would allow receivers to track to much lower C/N0 values and perhaps sail through even very strong solar radio bursts.
As we approach the peak of the next sunspot cycle in 2012, we can expect more solar radio bursts. Some forecasts peg the next peak at 30–50 percent stronger than the last one as measured by the fraction of the sun’s visible hemisphere with sunspot activity. Will future solar radio bursts have as dramatic an effect as the burst of December 6, 2006? Time will tell.
The FAA’s announcement (reported in March GPS World) that WAAS in the northeastern United States and eastern Canada may be significantly inhibited by relocation of WAAS-broadcasting satellite AOR-W before the new PanAmSat becomes fully operational in fall 2006 caused unease in some surveying organizations. Based on tests completed last year, before anyone knew that AOR-W would relocate to 142W longitude, these organizations replaced legacy GPS mapping units using post-processing and the Coast Guard NDGPS with high-performance WAAS-enabled mapping receivers.
The FAA notice doesn’t tell the full story, however. Two new WAAS broadcasting satellites launched last fall. PanAmSat (133W) began broadcasting in test mode with corrections full-time this March, and Telesat (107W) is scheduled to begin the same mode on or around April 1, 2006. The FAA announcement does not take into account either of these broadcasting satellites.
If these test signals are considered, there will be no degradation in WAAS visibility. In fact, users in the northeastern United States and eastern Canada will enjoy dual WAAS satellite coverage. WAAS satellite visibility in central and western North America has improved in the past 60 days with the new test signals and relocation of AOR-W.
However, the FAA won’t certify the accuracy/reliability of the new satellites until after extensive testing. Until then, non-aviation receivers may use the signals at their discretion —the same mode WAAS operated in prior to its July 2003 commissioning. Also, non-aviation WAAS receivers may not be configured to use the new test signals; check with the manufacturer.
— Eric Gakstatter, Editor of GPS World’s new Survey & Construction E-Newsletter
U.S. federal agencies, aircraft and avionics manufacturers, airlines, and research centers are brainstorming ways to prevent a repeat of the tragic events of September 11, in which highjacked aircraft were used as missiles. Under these new circumstances, options previously dismissed out of hand suddenly are attracting renewed attention.
One recurring proposal is to automate the landing of hijacked aircraft. In this scenario, a “dead-man switch” would allow the pilot to turn over navigational control to an on-board GPS-based autolanding system. The system would broadcast a mayday to air traffic control (ATC), search an on-board database for the nearest suitable airport, alert that airport, receive landing authorization, and land the aircraft there. During these operations, no one on board would be able to regain control of the aircraft. The pilot would be like an employee who, when confronted by a robber, does not have the combination to the company’s safe. No amount of violence on board would allow hijackers to use an aircraft as a missile against a target.
Technologically feasible. The strong consensus of airline, industry, and academic experts interviewed for this article is that the above scenario is technologically feasible. In fact, the autolanding technique has been amply demonstrated and at least one major avionics manufacturer is actively working on producing an emergency landing system.
The Federal Aviation Administration (FAA) is working on two GPS-based systems that could enable this sort of antihijacking capability: the Wide Area Augmentation System (WAAS) that will enable aircraft to reach the so-called Category 1 decision point in an approach to an airport, and the Local Area Augmentation System (LAAS) that would enable aircraft to reach the ground in zero visibility, known as a Category 3B landing. The agency plans for many airports to be equipped with LAAS transmitters eventually and will require WAAS/LAAS systems on commercial airliners.
Although both systems still await final certification, testing, and installation at U.S. airports, commercial airliners and military aircraft have already demonstrated fully automatic instrument approach and landing under Category 3B conditions.
Features
Although technologically feasible, operational considerations pose obstacles for implementing an automated emergency landing system. The following scenarios address some of these issues as well as technical features of such a system:
A “multiple key” arrangement could restore manual control with codes from the pilot, the co-pilot, and the ground-based ATC operators. Ground control would con- tribute its code only when absolute sure that the aircraft could not be used to attack a population center.
To protect it from being disabled, the system would require a hardened compartment not accessible from the cabin and an autonomous power source not controlled from the cockpit circuit breaker panel.
Prior to landing, the onboard system would notify ATC, which, in turn, would alert and re-route other aircraft as needed.
If the highjackers jammed the GPS signal, the system would put the plane in a holding pattern until it reacquired a clear signal. By refusing to turn off the jammer, terrorists could force the aircraft to run out of fuel and crash – but could not guide it to a target.
According to an industry source, the system should first put the aircraft in a holding pattern in any case, to give a chase plane time to reach it and visually monitor it. In the very unlikely case that the highjackers were able to regain control of the aircraft and aim it toward a target, the chase plane could challenge the aircraft, order it to land, and shoot it down if it did not comply.
The airport database would need to include data on possible flight path obstructions – terrain or tall buildings – so that the system could select a clear approach path. Avionics systems coming onto the market that are designed to prevent controlled flight into terrain essentially have this capability now.
The autolanding system would require permission from the ground to land on a particular runway. If permission were denied for any reason, the system would search its database for the next-best runway.
Cockpit philosophy. An airline pilot who is now an aide for the operations chief of a major airline reacted very negatively to the idea of an emergency autolanding system that could not be disengaged by the pilot. Any system that restricts the crew’s options, he said, clashes with a key tenet of “cockpit philosophy”: to keep the pilot in charge and never relinquish control of an aircraft completely to automation.
An emergency autoland system also conflicts with a basic principle of aeronautical engineering – namely, that an aircraft should have multiple, redundant ways to control it.
However, in extreme emergencies, the alternative may warrant overriding such concerns, according to Bradford Parkinson, a professor emeritus at Stanford University’s School of Engineering who first proposed fully automated cargo planes years ago. He points out that, although an antihijacking system used routinely would have to be extremely reliable, when the alternative is a 100 percent probability of death for all aboard, “Boy, that sure changes the equation in a flash.”
Further reading: “Soft Landings: Navy Proves Hands-Off Touchdown,” by Matteo Luccio and Glenn Colby, GPS World, August 2001.
