Air Traffic Control Newsletter #159

Air Traffic Control Reform Newsletter

Air Traffic Control Newsletter #159

Revisiting the SFO near-miss and FAA's "fixes" for detecting taxiway landings, GPS landing systems, and more.

In this issue:

FAA “Fixes” for Taxiway Landing Detection

In the wake of the July 7, 2017 near-disaster at San Francisco International Airport (SFO), when an Air Canada A320 came within 45 feet of landing on top of three jetliners waiting for takeoff on a taxiway next to the runway, aviation media this fall have included several articles about fixes to this problem. FAA has announced that a modification is under way at the 35 major airports equipped with the ASDE-X surface-detection system. And two long-time FAA suppliers—Honeywell and Raytheon—have proposed other fixes.

Aviation Daily published the official FAA version of the ASDE-X fix on Sept. 27. It said the agency had “put together a ‘tiger team’ to explore whether the ASDE-X system installed at 35 major airports could be modified to help detect when aircraft are about to land on taxiways.” And it said that “FAA determined that technological advances meant that FAA could add so-called ‘Taxiway Arrival Prediction’ capabilities to ASDE-X.” It further reported that the agency had tested the modification at Seattle (site of an earlier taxiway landing incident) in January and had completed modification of ASDE-X at SEA in May. FAA informed the National Transportation Safety Board on Sept. 13 that evaluations had been completed at four more airports (Atlanta, Boston, Charlotte, and Hartford) and that the modifications would be in place at all 35 ASDE-X airports by Sept. 30, 2020. (A Nov. 15 follow-up story reported that the warning capability was now in operation at those four airports.)

This account is somewhat at odds with the facts. As I reported in this newsletter’s issue number 146 (September 2017), ASDE-X, as created and installed by Saab Sensis, had taxiway-landing detection capability built in, capable of detecting an errant plane’s approach up to three-quarters of a mile away. This capability was in the original 2009 requirements document and is deployed on ASDE-X in Europe. Knowing of this led NTSB in 2009 to recommend that FAA review the company’s information and, if valid, to turn that capability on. But then-Administrator Randy Babbitt rejected the recommendation to even study doing this, which NTSB deemed an “unacceptable” response.

To get further insight into this situation, last fall I interviewed an ATC engineer who had worked on ASDE-X and its predecessor, AMASS. Here is what he explained to me:

“Taxiway warning capability was an original AMASS requirement, but when the early system was turned on, it produced a false-alarm rate that tower controllers found unacceptable. The addition of multilateration capability, deployed as part of ASDE-X in 2003-2011, added increased accuracy, making the “Arrival to Taxiway Warning” feature functional, but FAA still refused to turn it on. In the wake of the SFO near-disaster, FAA now says it is working on software upgrades to ASDE-X for this purpose, and may begin testing within a few months . . . . In fact, no software upgrade is needed; the warning module simply needs to be activated and adapted to each airport.”

What FAA is actually doing, and why it will take two more years to have this vital warning feature in operation at all 35 ASDE-X airports, are questions the DOT Inspector General’s Office might well look into.

Meanwhile, always on the lookout for new business, two long-time FAA contractors are proposing additional technologies to deal with this problem. Raytheon, which has nearly finished the years-late, far-over-budget STARS program, now re-named TAMR (deploying a replacement for the aging CARTS automation system for TRACONs and associated towers), is proposing an add-on to that system called Arrival Runway Verification (ARV). It would monitor the flight track of aircraft heading toward a runway and generate a “wrong runway” alert if the plane is lined up for a different runway than it has been assigned to. Raytheon points out that ASDE-X is in operation only at 35 major airports, whereas STARS will be available for airports large and small. It also claims that ARV would provide up to two minutes warning (which my engineer friend says is not believable), as opposed to 30 seconds with the warning feature in ASDE-X.

Honeywell has finished initial testing of a cockpit system to warn pilots of possible conflicts on the airport surface. Called SURF 1A, it uses ADS-B data to alert pilots to possible conflicts with other aircraft, in the air or on the ground (e.g., on a runway or taxiway). The project is being developed by Honeywell in partnership with Airbus, Dassault, and Eurocontrol. This would likely be an airline expense, not an FAA expense.

I asked my original engineer informant for any comments on these developments. He emailed back that, “There are these three different initiatives going forward in three [FAA] organizations with zero oversight at the top. ATO has no single systems engineering organization looking at solution sets.” He also noted that “the FAA engineering team in Oklahoma City likes to ‘reinvent their own stuff,’ putting in software and modifications that are not needed—repackaging what already exists and claiming new stuff.” He also reiterated that ASDE-X installations in Europe have long included the Arrival to Taxiway Warning capability, without encountering any problems.

» return to top

Space-Based ADS-B Moves Toward Operation in Europe and USA

Efforts are under way in both Europe and the United States to bring Aireon’s space-based ADS-B surveillance to operational status. The European Aviation Safety Agency (EASA) is far along on certification of the service for use by air navigation service providers (ANSPs) in Europe, five of which are already planning to make use of the service. EASA certification will be part of the safety case that each ANSP will carry out, explaining how space-based ADS-B surveillance data will be integrated into the ANSP’s operations.

More recently, the FAA has begun a two-year operational evaluation of space-based ADS-B, as discussed in detail in the Oct.15-28 issue of Aviation Week. Since the FAA is responsible for air traffic control in large amounts of oceanic airspace (including much of the Pacific, parts of the North Atlantic, and most of the Caribbean), the introduction of ADS-B in that airspace is expected to lead to reductions in the very large lateral and longitudinal spacing (30 nm for each) between aircraft required by “procedural separation” in airspace where radar surveillance is absent.

As a refresher on how Aireon’s system will work, its ADS-B package is included on the entire constellation of new IridiumNext satellites. ADS-B/Out signals from equipped aircraft are received by the nearest satellite (providing position, velocity, and identification data), and the satellite transmits that information to Aireon’s ground segment for processing. That information is then transmitted to the relevant ANSP’s automation platform, to appear on controllers’ screens.

FAA’s evaluation will begin with Caribbean airspace, where aircraft are tracked from the oceanic control center in Miami. Space-based ADS-B data will be fed into that center’s ERAM system, a process already completed for the ERAM system at FAA’s William J. Hughes Technical Center in Atlantic City. The Caribbean is an easier oceanic environment because communications between controllers and pilots can use VHF radio signals relayed by island-based ground stations. But in oceanic environments far removed from land, the only existing controller/pilot communications are provided by ADS-C, which only sends messages at specific intervals. That is not sufficient in cases where quick back-and-forth communication is needed, so some form of satellite-based two-way communications link is being looked into by FAA, Aviation Week reports. Nav Canada is also considering the implementation of satellite voice communications for this purpose.

The Caribbean effort is expected to involve several U.S. airlines by 2020. And later steps will include integrating space-based ADS-B data feeds into the ATOP automation system used at the New York, Oakland, and Anchorage centers responsible for North Atlantic and Pacific oceanic airspace. This multi-year evaluation is leading up to an FAA funding decision, Acting Administrator Dan Elwell told the Air Traffic Control Association’s annual meeting in October. Elwell also told that attendees that he sees space-based ADS-B as “a natural evolution of the technology development that we have seen over the decades,” and that he sees it as “the wave of air traffic control in the future.”

» return to top

Inspector General Finds Continued ERAM Shortcomings

FAA completed installation and testing of the new ERAM automation platform at all 20 en-route centers in March 2015. The process took over four years longer than originally planned, and at $2.1 billion, it was hundreds of millions of dollars over budget. But it is a major improvement over the obsolete HOST computer system that it replaced.

Unfortunately, in 2014 two major failures of ERAM took place, which shut down all operations at the Los Angeles and Washington centers—a condition known as “ATC Zero”—resulting in hundreds of flight cancellations and thousands of flight delays. Both problems were attributed to software problems that were identified and fixed before ERAM was declared fully operational in 2015.

But a new report from the DOT Office of Inspector General (OIG) identified five other ERAM failures between 2014 and 2016, at the Denver, Memphis, and Washington centers. None qualified as ATC Zero, but they were serious enough to be classed as ATC Alerts. While all five of these have been corrected, mostly via software patches, OIG found several additional problems that remain to be remedied. The report, “FAA Has Taken Steps to Address ERAM Outages, but Some Vulnerabilities Remain” (Report AV2019004, Nov. 7, 2018), identifies the additional problems and reviews what FAA has and hasn’t done about them.

One cause of the ATC Zero at Los Angeles center was human error. In that incident, a controller entered a military jet’s altitude as 7,000 ft. rather than the actual 60,000 ft. That made the system react as if chaos was about to occur for all other traffic in that flight region. FAA issued new instructions about entering altitude data for high-altitude flights.

A more serious problem was identified as inadequate testing of contingency plans. Guidelines from the National Institute of Standards & Technology call for having functioning backup systems, annual testing, and detailed procedures for staff on how to recover a system in emergencies. While ERAM has a backup system (called EBUS), there is no annual testing; hence, controllers “are not gaining experience in reacting to and recovering from outages” like those ERAM has experienced. Moreover, FAA “does not adequately provide [annual] training for staff on ERAM’s contingency plan.” Based on a previous OIG report on this subject, FAA told OIG that it is “developing long-term requirements for contingency planning and controller training,” but the report notes that “it is unclear when these new requirements will be implemented.”

The third problem highlighted by the OIG report is ERAM’s questionable backup capability. The original design did not include a backup system, because ERAM already had dual channels, in which one could take over from the other. However, the Los Angeles center ATA Zero involved both channels failing at the same time. FAA therefore “temporarily” retained the old HOST backup system (EBUS). Upcoming modifications to ERAM to accommodate new hardware and to make it compatible with upcoming NextGen systems means EBUS will not be capable of backing up the revamped ERAM, so FAA plans to retire EBUS as early as April 2019. OIG concludes that “The lack of sufficient backup capabilities could increase ERAM’s vulnerability in the event of future unanticipated incidents.” Therefore, the report recommends that FAA “determine what [future] backup capability is required for ERAM and implement that capability.”

After reviewing a draft of the report, FAA only partially agreed with its recommendations, so OIG is considering them “resolved but open pending completion of planned actions.” In other words, this story may yet have further chapters.

» return to top

Data Comm Under Budget, Ahead of Schedule

Though begun about a decade after controller-pilot data communication became operational in Canada, FAA and Harris Corporation have made great progress on implementing Data Comm, which permits controllers and pilots to send text messages rather than using noisy and often-congested voice channels. The first phase of the program called for installing Data Comm in 55 of the system’s busiest airport control towers. Harris completed that by the end of 2016, two years ahead of schedule. And since it also finished under budget, FAA spent the savings to equip five more towers, the last of which became operational Aug. 21, 2018. This initial phase of Data Comm allows controllers to send pre-departure clearances and any last-minute updates direct to flight decks as planes await pushback, saving precious minutes over previous voice procedures (and potentially avoiding mis-hearings and mis-entries of data into the plane’s flight management system).

Phase 2, adding Data Comm to all 20 domestic en-route centers, is now under way, with a target date for completion by the end of 2019, several years earlier than previously planned. The first three centers being equipped are Indianapolis, Kansas City, and Memphis. Initial messages planned for text delivery include changing to a different radio frequency, assigning altitudes and altimeter settings, and reroute information, as well as pilot-initiated messages such as requesting direct routings. Additional message sets are being developed for the centers, to be implemented through about 2025. After that will come “advanced services,” such as dynamic RNP, advanced interval management and 4-D trajectories.

FAA reports that 4,900 US aircraft were Data Comm-equipped as of August. That includes the majority of the major airlines’ fleets (passenger and cargo) plus about 1,000 general aviation and business jet aircraft. These planes use existing FANS 1/A data link communications that have been in use since the 1990s on oceanic routes to provide controllers with updated position reports via satellite. This operates over VHF Data Link Mode 2 (VDL2) radios. But by 2030 the communications protocol will shift to Aeronautical Telecommunications Network Baseline 2 (ATNB2), still using VDL2 radios, to be compatible with Europe’s data link requirements. Suppliers of the VDL2 radios—Rockwell Collins and SITA—are upgrading their radio ground station networks to handle the increased communications volume.

It’s a pleasure to report on a NextGen program that is under budget and ahead of schedule.

» return to top

An Expanded Case for GPS-based Landing Systems
By Capt. Tom Imrich

Editor’s Note: Last issue’s article on the advent of GPS-based landing systems (GLS) capable of Category III (Cat III) precision generated a number of emails questioning the case made in that article. In response, I have asked a highly qualified colleague to provide the response below.

GLS/GBAS is a highly capable, efficient, and economical navigation system that will increasingly serve as the central element of the C-N-S (Communication-Navigation-Surveillance) triad for takeoff and landing, in conjunction with Required Navigation Performance (RNP). Anybody who has flown it, especially in stressful weather conditions, can attest to its amazing capability. But GLS/GBAS not only supports all-weather operations (e.g., Cat III requirements), but can also define an accurate path for safe takeoff and landing, for every operation, for all types of vehicles, not just jet transports, even in clear VFR weather. Here is a brief recap, explaining why GLS/GBAS is needed globally:

  • GLS provides an accurate, reliable, high-integrity path over the entire runway, allowing for safe “proportional, integral, derivative” (PID) flight control for every takeoff and landing operation.
  • One GBAS facility can serve multiple runways (i.e., all runways at an airport, even some nearby airports).
  • GLS does not have any “critical areas” where it is vulnerable to interference, as does ILS.
  • GLS is far less prone to outages, like those that plague ILS.
  • GLS/GBAS can be substantially less costly than ILS (e.g., easier siting, no extensive site preparation).
  • GLS is easier to maintain than ILS and needs little or no recurring flight inspection.
  • GBAS can help improve other navigation capabilities, such as for FMS-based RNP navigation.

Now let’s address some misconceptions about GLS’s role, status, capability, integrity, cost, and benefits.

Misconception #1: Present GLS-equipped jets aren’t yet good enough for Category III. The real issue here is not the capability or integrity of the present GLS airborne system or GBAS ground facility. Instead, the problem is the use of faulty criteria by some aviation safety regulators, such as FAA. Present criteria for GLS are substantially “over-specified” and unnecessarily stringent, far more than what is required of ILS. From the outset, current GLS-equipped jets such as B737NGs and B787s were designed to use GLS to operate to Cat III minima. These jets are substantially better in all three components of accuracy, integrity, and availability than earlier jets approved for Cat III operations (such as the L1011, A300, or B727). The new jets each fully meet provisions of criteria such as FAA AC120-28D (for Cat III), even based on available GLS/GBAS ground facilities. The FAA approval statements for those jets typically did not include “demonstrated capability” for Cat II and Cat III for political, not technical, reasons. Hundreds of already delivered GLS-equipped transport jets could safely be using Cat III minima today, with presently installed GBAS facilities, if only more-appropriate criteria were applied.

Misconception #2: Present GLS/GBAS ground facilities aren’t yet good enough for Cat III. Current GLS is typically being asked to meet unrealistic and excessive standards by FAA and some other regulators, well beyond criteria that even the latest ILSs could realistically meet. Currently deployed GBAS facilities are entirely adequate from a perspective of accuracy, integrity, and availability, to support limited Cat III operations, if realistic criteria are credited for both the GBAS facility and the aircraft using that facility.  This is because the combination of monitoring protection in the GBAS facility and the design of the aircraft flight guidance system provides for the needed accuracy, integrity, and availability. As an example, GLS has been demonstrated to provide a safe and reliable autoland to a complete stop even following a complete GBAS failure as the jet descends below 200 ft. In comparison, ILS not only has frequent beam interference, but it even degrades on its own in some types of weather conditions. GLS/GBAS facilities are largely immune to these kinds of failures. Present GBAS facilities are already better in terms of accuracy, integrity, and availability than most of the early ILSs used to support Cat III operations, back in the ‘70s through ‘90s.

Misconception #3: GLS Paths are rigid and inflexible, and can’t be used by controllers to meter and space (M&S) aircraft efficiently, thus decreasing airport capacity. Admittedly, this is a complex subject requiring consideration of many technical aspects. In the end, the claim is also misleading. The short answer is that GLS/GBAS is intended to be used along with flexible dynamic 4D Required Navigation Performance (RNP)-based curved and segmented paths, and not just used alone, as a long fixed “straight in” final approach landing path. If used properly with RNP, GLS can help facilitate M&S, to increase airport capacity, such as by better-using runways, and by using multiple runways simultaneously, while still avoiding weather and noise-sensitive areas. However, the observation that FAA currently doesn’t have all the needed decision support tools to currently benefit from the use of RNP and GLS has merit. This has been known for decades, and FAA systems for NextGen, such as STARS, need to be expeditiously upgraded to take advantage of dynamic flexible 4D paths, with GLS/GBAS.

Misconception #4: GLS/GBAS facilities will be “high cost” compared to ILS. GLS/GBAS facilities should actually be much lower-cost than ILS. Since GLS simultaneously covers multiple runways and multiple directions and is simple to build and install, GLS/GBAS is likely to be much lower cost than ILS. If fairly assessed, GLS/GBAS has a clear advantage for initial acquisition, installation, operation, flight inspection, maintenance, and life-cycle cost, compared to any ILS. A GBAS facility at its heart is simply a GPS receiver, hooked to a VHF radio that sends up GPS satellite “state vector” corrections to any nearby aircraft that tunes in. Current GLS/GBAS cost numbers are often substantially inflated due to use of technically inappropriate criteria, inappropriate defined market size, incorrect assumptions about all-weather operations, lack of competition among GBAS facility sources, or faulty assumptions used in cost/benefit studies. GLS/GBAS costs estimates could be an order of magnitude less than some of the present high estimates.

In this modern era of extremely capable avionics systems and displays, with the incredible power of modern electronics (e.g., cell phones and iPads), it should be possible for every air vehicle, from drones to jet transports, to have a suitable, reliable, precision-based path defined, with integrity, for every takeoff and landing, at every runway. With such a path available, and flight path control to maintain it, there is simply no reason any more for accidents or incidents such as at Halifax, Nova Scotia (Air Canada AC624), or San Francisco (Asiana 214 and Air Canada 759). Every new aircraft avionics system built should now include GLS/GBAS capability, in aircraft ranging from drones, to air transport, to military aircraft. For air navigation service providers (ANSPs), GLS/GBAS needs to be deployed at every jet transport airport globally, to economically provide an accurate high integrity path to and from every runway used.

Capt. Tom Imrich previously served as Chief Research Pilot and B747 Senior Engineering Pilot for Boeing. Prior to Boeing, he held various positions at FAA, where he was a principal author of numerous All-Weather Operations criteria and rules, including FAA Advisory Circulars for Cat II and Cat III landing. While at FAA, he was directly involved with nearly all modern jet transport aircraft Cat II and Cat III certification. At Boeing, he was Co-Captain for the B747-8’s “First Flight” and was a primary DER/AR test pilot for GLS development and FAA certification on both the B737NG and B747-8, as well as helping with GLS on the B787. Capt. Imrich retired from FAA in 2001 and from Boeing in 2012.

» return to top

News Notes

Final Iridium Launch Set for December 30.
The eighth and final launch of 10 more IridiumNext satellites is set for Dec. 30. As before the launch vehicle will be a SpaceX Falcon 9, and the launch site will be Vandenberg Air Force Base. With the full constellation of satellites in operation early next year, Aireon will begin offering global space-based ADS-B surveillance to air navigation service provider (ANSP) subscribers to the service. Iridium itself will be offering an array of other communications services via the new constellation.

Nav Canada Implements New ICAO Separation Standard.
Canada’s self-supporting ATC corporation announced on Nov. 22 that it is the first ANSP to implement ICAO’s new “Established on RNP-AR (EoR)” standard. It allows aircraft that are qualified for RNP-AR to make simultaneous landings on closely-spaced parallel runways. The new standard has been implemented at Calgary International Airport (YPC). More than 40 percent of airliners using YPC are already RNP-AR equipped. Between the use of continuous descents and curved approaches under RNP-AR, Nav Canada estimates airlines will achieve fuel savings, as well as 2,500 metric tons less greenhouse gas emissions in the first year.

Aireon/FlightAware Flight Tracking Available Now.
Aviation Daily
reported (Nov. 6) that Aireon and FlightAware have activated their new GlobalBeacon flight tracking service. It is offered on a subscription basis to airlines and business jet operators to provide minute-by-minute tracking of aircraft, particularly in polar and oceanic airspace where there is no radar coverage. GlobalBeacon is one way that aircraft operators can comply with a new ICAO tracking recommendation (Global Aeronautical Distress Safety System) that became effective Nov. 8. The initial requirement is for updates no less than every 15 minutes, but as of 2021 this changes to once per minute, which GlobalBeacon already meets.

FAA Unveils New PBN Routes to Florida and Caribbean.
On Nov. 8, FAA implemented 55 performance-based navigation (PBN) routes between airports in the Northeast and the major airports in Florida and the Caribbean. The project is part of FAA’s South Florida Metroplex program. The addition of the new routes brings the total to 316 PBN routes in the United States.

First Remote Tower in South America Goes to Argentina.
Buenos Aires’ Ezeiza International Airport will be the site of the first remote tower in South America. It will be implemented by Frequentis, as part of its SmartVision system, developed with Rheinmetall Defense of Germany. The remote tower will supplement the existing conventional tower during a major expansion of the airport’s terminal capacity. The expansion also includes a new conventional tower, to which the remote tower equipment will be relocated once that tower is ready to begin operations.

Saab/LFV Joint Venture Now Controlling Traffic at Two Airports.
The ANSP of Sweden—LFV—and aerospace company Saab recently created a joint venture company called Saab Digital Air Traffic Solutions to work together on remote towers (which some are calling digital towers). The company has recently taken over operation of local air traffic control at two Swedish airports: Sundsvall (where the Saab remote tower center is located) and Ornskoldsvik, which is controlled from the center in Sundsvall. The initial agreement under which SDATS manages traffic at the two airports is for six years.

Congratulations to Linda Hall Daschle.
The 2018 Glen A. Gilbert Memorial Award was presented to former FAA Acting Administrator Linda Hall Daschle. Among other accomplishments, she made the decision to terminate FAA’s out-of-control Advanced Automation System program. She is on the board today of space-based ADS-B provider Aireon. The Gilbert award is given each year by the Air Traffic Control Association to someone who has made outstanding contributions to air traffic control. It is named after Glen Gilbert, the first controller employed by the Bureau of Air Commerce when it took over the fledgling ATC system developed by airline user co-op ARINC. In his retirement years, Gilbert produced a two-volume study calling for ATC to be divested from FAA to a “Comsat-type corporation” funded by user fees and taxes.

Senate Takes Step Toward GPS Backup.
In November the Senate passed the National Timing Security and Resilience Act of 2018, which requires the DOT Secretary to establish a terrestrial national timing signal to provide a backup for GPS timing. The bill is included as part of the US Coast Guard Authorization Act, a presumably must-pass bill.

Aireon Moves into Brazilian ATC.
Space-based ADS-B provider Aireon has signed a memorandum of understanding with Atech Negocios em Techologias, a Brazilian developer of ATC automation platforms. Under the MOU, Atech will evaluate the incorporation of space-based ADS-B surveillance data into its Sagitario system, which is operational in five Brazilian area control centers (ACCs).

» return to top

Quotable Quotes

“[Air traffic management] appears to be on the cusp of a bright new digital future. . . in which remote towers and virtual centers offer new ways of providing air traffic control. Central to this is the idea of services. Rather than vertically integrating all elements of ATC into monolithic area control centers, various data services will be provided using system wide information management (SWIM). . . . The ability of a network of virtual centers to be reconfigured in terms of airspace designations to cope with the loss of an ACC, or perhaps even to cope with significant shifts in air traffic flows as occur with geopolitical events, such as the current situation in Eastern Europe, is going to be a huge step change in the resilience of air traffic management.”
—Paul Ravenhill, “Loud & Clear,” Air Traffic Management, Issue 3, 2018

“Your September article regarding tower radar displays and remote towers touches on a bigger issue. . . . I understand that most, if not all, of the FAA’s roughly 90 VFR towers (Class D) have FAA-furnished STARS tower displays. However, only a fraction of the 263 federal contract towers are similarly equipped. (This may challenge your statement that ‘federal contract towers have Dbrite scopes.’) I’ve asked many levels of FAA to explain their policy (e.g., traffic levels, surrounding airspace), but continue to believe one doesn’t exist. One wonders why the tower at Frederick, MD, which experienced a mid-air near the traffic pattern a few years ago, still does not have a tower display. I’ve also visited two non-federal control towers that have STARS displays; it’s unclear who pays for them.”
—Email to Robert Poole, Oct. 26, 2018, name withheld by request

“I’m extremely excited about what Data Comm can afford the national airspace system in terms of safety and efficiency. This is about execution and agility in that execution. What we are counting on is the simplification and standardization of tens of thousands of transactions a day in the system.”
—Richard Dalton, Southwest Airlines, in Bill Carey, “Data Comm Delivers,” Aviation Week, Oct. 1-14, 2018

» return to top