Slashdot Mirror
FAA Pushes Air Traffic Control Systems Into Service
An anonymous reader points us to this AP story about the FAA forcing new air traffic control systems into service, over the objections of technicians and air traffic controllers. The Transportation Department's Inspector General notes that many critical bugs remain unfixed. We reviewed a book that discussed the lessons to be learned from software engineering projects; and we had a recent story about Great Britain having all sorts of problems with their new air traffic control software.
for the CLIT
The
Even a buggy air-traffic control system is better than the one they have now. Every 3 days or so, the radar screens drop horizontal so that air traffic controllers can keep track of their dots while the computer goes down and then comes back up.
This is not a pretty sight. Each ATC can have 30+ planes on their screen, and when the computer goes down, they screen drops horizontal, the ATC whips out little flags with numbers on them and keeps them on top of the now anonymous dots.
So I think ANY improvement is a good thing.
On the blacker side: The bugs themselves could be a good thing. Maybe one of these "bugs" will misreport the location of things like the Sears Tower, or the Capitol Building and a hijacked plane will slam into "Al's Meats and More" instead of the intended target.... (yes, I'm still REALLY angry about the Pentagon/WTC/Penn. terrorist attacks)
The Dopester
"Yes, I'm a Karma Whore, but I'm doing it to pay my way through school."
It lists a few problems, but doesn't say if it could be the result of user error. We all know how users always say the system crashed when it's really their mistake
Does it matter? I know I wouldn't care why the system crashed if I was in the plane. I'd just want to get home alive. If the system crashes because of user error, then that is a serious flaw.
t'nera semordnilap
Such a troll....
Imagine: the buggy (and needless to say proprietary) flight control software is installed. Two months later, plane crashes are at an all time high. The FAA is in an uproar, the media is clamoring for a solution, America is in turmoil.
I'm just not seeing this happen. First of all, the old systems sucked just as bad, and they're keeping them around for backups. Secondly, the biggest danger here is mid-air collisions, and modern aircraft have at least good enough proximity alarms to avoid a problem. I would be more worried about smaller private craft than commercial airliners.
What's that? It's the phone. Who is it? Someone named Linux Torvalds...says he has a solution to our problems.
Wrong. This is serious life or death stuff. Even with real time extensions and a VM that doesn't change every 30 days, Linux is neither designed nor intended to handle high load real time situations. Open source software just simply isn't hardened enough to do anything approaching the complexity of a major air traffic control system.
Don't get me wrong - Linux is a good operating system, and its fun as hell to use and hack on, but a good engineer knows that there is a right tool for every job, and Linux just isn't a legimate choice for a high priority real time system like the ATC.
Do you have Linux and a DotPal? Click here now!
Avweb has also had some interesting articles about England's experiement with new ATC systems.
sPh
Raytheon Canada has developed Canadian Automated Air Traffic System (CAATS) and Military Automated Air Traffic System (MAATS) for a Nav canada contract and are currently selling it to other countries as well. The software works really well (I have seen it firsthand). It was developed in Richmond, BC.
Of course, the air traffic controller's union is upset about this. Most air traffic controllers have a very short career (burn-out) and don't want to have to learn a new way of doing things.
Please read:
The ATC / Control Tower is never responsible for controlling the aircraft, nor actually guiding aircraft between hills/etc. Even in a busy place like the San Francisco Bay, the ATC advises of traffic and coordinates inbound and outbound traffic lanes to keep traffic well spaced. ATC typical instructions, even with hills very nearby (1nm) usually consist of a destination landmark or vector, and an altitude.
Not to rant or anything, but from what I understand, the current system is incredibly out of date. It is important, not only to update the ATC gear, but that the FAA institude a complete overhaul of avionics. It is now possible with less than $1000 of gear to have in-flight collision detectors based on GPS, but this is uncommon on private aircraft costing $100k+.
Just my 2 cents.
At the point that they're that close, it's all on the pilots as to where they land. The controller will tell them to circle until it's their turn to land.
In busier airports at night (think LAX, DFW, ATL) you can see the planes lined up 2 minutes apart. That's what the controllers do. The pilots are responsible for centering themselves on the runway, and landing without crashing into the ground.
Whoever stated that signature sizes should be limited to one hundred and twenty characters can just go ahead and kiss my
I am a private pilot. Even if ATC completely shut down aircraft would not start falling out of the sky. Airplanes are flown by pilots who are trained to operate the aircraft completely independently from air traffic control.
Think of air-traffic control as stop-lights for automobiles; when the stop lights go down, do traffic accidents start happening? No, you just get a little less efficient traffic flow (in some cases it gets more efficent...). Drivers know how to take turns just like they do at stop signs.
Analagously, pilots know how to take turns and fly safe just like they do at 90% of the airports in the world that don't have 24x7 air traffic control.
If the street lights start malfunctioning and giving wacky signals, the hazard of accidents might go up, but would not neccessariy lead to catastrophe.
The ananology for aicraft is even stronger: if an ATC controller went mad and decided to purposefully cause an accident, he probably would not succeed since he would have to fool two pilots who are trained to be wary of ATCs command and to overrride them when they are in error.
Bottom line: airplanes are flown by pilots, not traffic controllers, so breathe easy.
Usually, the connection is not hardwired, but transmitted from one site to another. I'm an IT Manager at a regional airport, and that's the system we use. We have a large Air Force base only 10 minutes away, and they got expensive upgrades before we did, so they basically just transmit their ATC data on over to us via microwave. This is not unusual. This kind of resource sharing is pretty common between military and civilian installations. We have have an Air National Guard squadron here as well, and we also share resources with them (firefighting in particular). It just makes more sense that way.
As for the new ATC system, I think any kind of mission critical system should be Unix based. As much as I like Linux, I'd still feel better with a QNX type system that goes down perhaps once a year or so. You have to wonder what the procurement people are thinking sometimes.....
Life is hard, and the world is cruel
Baseline magazine, a periodical dedicated to IT management, covered the new FAA STARS system from an IT project management point of view in a recent issue.
You can view the case study on-line, or download the PDF file.
I found it to be a very good in-depth article delving into many of the issues surrounding such a massive IT project.
The AvWeb article cited above (which is written by Don Brown, Facility Safety Representative at ZTL) also talks about FAA wanting to do away with primary radar altogether. Fucking morons. There are still plenty of airplanes flying around with no electrical systems, which means no transponders.
Unlimited growth == Cancer.
(Before I get started with my critique of the article and my take on the STARS issue, for those of you more interested in the fact that Sun/Solaris is at the heart of STARS, skip to the bottom of the page)
I've been an Air Traffic Control Radar Tech for the better part of 20 years and, after watching/reading years worth of inaccurate FAA Technology reporting (such as this CNN piece) I'm convinced that aviation journalists are, for the most part, clueless about the technology that they report on. Combine ignorance of that calibre with the natural alarmist tendencies of journalists and editors seeking incresed circulation/viewership and you end up with pieces like this one. Alternatively, and even more unpalatable, it might just be that all journalists are mere dilettantes, and actually have the barest grasp of the issues about which they write. Perhaps, as radar/air traffic control is my field of expertise, I'm only accutely aware of their shortcomings in that field, whereas they may be just as ignorant in many, or all, others. I hope that's not the case, but reporting that's as slipshod as this makes one wonder.... The fourth estate is just as prone to error and exaggeration as the other three.
Obviously, there are problems with STARS, just as the DOT IG report describes. There are problems with ALL new major FAA systems. I've been involved with the ASR-11 program, off and on, for several years now. The ASR-11 is a short range (airport) radar that, like STARS, is a Raytheon product and is currently undergoing a variety of tests to assess it's suitability for inclusion into the Air Traffic Control system. The STARS and ASR-11 sagas have similarites: both have been undergoing testing and some forms of development for years. A portion of the agony involved in equipment acceptance is rooted in the diverging interests of the vendor and the FAA. The vendor claims that the system will perform in such and such a manner, and it's up to the FAA to verify that their claim is accurate. If the claim cannot be verified, then a fix must be proposed, then agreed upon, then implemented, and then verified. Of course, the vendor and the FAA will interpret various aspects of the contract differently, and therefore problem resolution can, and apparently often does, involve disputes about funding: who pays for the resolution? does the FAA cough up more cash or does the vendor eat the cost? I don't use the word 'agony' as hyperbole: it most certaily is agonising for both the vendor and the FAA. However, don't be too quick to blame beuracrats. The FAA is attempting to walk a very fine line: save costs where possible, and therefore give the taxpayer better bang for the buck, while ensuring that the system in question is as safe, and reliable, as possible. Of course, altruism isn't the only motivating factor: I imagine that Congressional oversight certainly helps, particularly when it comes to bang-for-the-buck considerations. However, I genuinely believe that engineering/testing personel, system maintainers, and air traffic controllers are supremely interested in the safety of the flying public, and act accordingly.
Now on to a critique of the CNN piece:
"The only STARS system now in use, in El Paso, Texas, has been plagued with problems, according to.....the Professional Airways Systems Specialists, the union that represents the FAA employees who certify and maintain air traffic control equipment."
My experience with the ASR-11 project has convinced me that the Technician's Union, Professional Airways Systems Specialists (PASS), doesn't give much of a shit about truly relevant equipment funtionality issues. I don't doubt that their assessment of STARS suffers from the same self-interested myopia. As I've heard it told, a Union's involvement in an early round of STARS testing turned into a fiasco, and a potentially significant opportunity was squandered. I've also heard that the union has learned a bit from the experience and that, perhaps, future Union involvement in STARS testing will be more productive. I have direct knowledge of some ridiculous Union demands vis-a-vis the ASR-11.
Unions are a more than a mild source of irritation to me, for a variety of reasons the reader could likely care less about. However, there is one aspect of unions that is crucial to this and other stories involving unions: a union, even one which whose membership comprises a fraction of the 'baragining unit' employees (those who could be in the union if they chose to be) is the sole representative of that group of employees and management looks to the union for all things to do with the employees, and seeks union write-off of all employee-related matters. Journalists follow the same pattern: they spout whatever line the union gives them as if the union actually, rather than technically, spoke for all the employees. If you follow aviation reporting you will see this proven true time and time again.
An aside: my opinion of the Union has nothing to do with my opinion of the average FAA technician. Anyone that's been exposed to unions understands that the official union position often bears no resemblance to the employee's position. FAA technicians are highly trained and, generally, highly motivated, and appreciate the serious nature of their profession.
"The old system remained in place as a backup, because "tower managers stated controllers were not comfortable relying solely on STARS,"....
Now this really turns my stomach....the fact that the 'old system' remains in place is somehow condemnation of the new system (STARS). Of course the old system remains in place as a backup: it would be grossly negligent to uneccesarily remove it while testing on the new system continues. What's so stomach-churning is that this hypocritical journalist, who obviously has a minimal grasp of the complexity of FAA equipment and the air traffic control system, would very likely be the among the first to accuse the FAA of negligence if the 'old system' was uneccesarily discarded and a failure of the new system resulted in crashed planes and mangled bodies. Look, the FAA KNOWS that, while the 'old system' is technically, well, OLD, it's tried and true and, therefore, safe. Obviously they're going to want to retain it as a backup, especially considering the birth pangs that STARS is experiencing. Keep in mind that STARS is not merely a new hardware backbone: it's a completely new interface as well, so it's new to both Air Traffic Controllers and Maintenance personel. Air Traffic Controllers take their responsibility to the flying public VERY seriously, and they're almost always, if not always, going to err on the side of caution. Any one that flies should appreciate this fact.
"Union vice president Tom Brantley said the radar doesn't always work, and it may require several minutes before controllers realize the problem. In addition, he said, the system has failed several diagnostic tests."
I don't know what the hell this means: STARS isn't a radar, of course, so I assume that he must be referring to the radar/s that feed the STARS. I have heard that there is an issue with lag under certain unique and rare circumstances. Those issues will most certainy be resolved prior to acceptance, or at least examined for validity. More to the point, this sentance is a perfect example of a cursory treatment of a very complex matter by someone that obviously has no idea what he's talking about (I mean the journalist, not the Union VP). It's confusing and meaningless, rather than informative and clarifying. Sloppy reporting, at best.
Now, back to a subject perhaps more interesting to the average geek: STARS systems are based on Sun/Solaris boxes, and LOTS of them. STARS, and other imminent and existing FAA systems, such as the ASR-11 (an airport/short-range radar), the WSP (Weather System Processor), all use Sun boxes. I believe that, between the Department of Defense (DOD) and the FAA there are going to be hundreds of STARS, and a couple of hundred ASR-11s, and over the next 5 to 15 years many hundreds, if not thousands, of technicians will receive various forms of Unix training. For several years the FAA has offered a three-week instructor-led Unix course and, I believe, requires this course as a prerequisite for those technicians who will be taking the STARS and ASR-11 Maintenance courses, among others. The course is based on Redhat and PCs, rather then Ultra or Sunblade and Solaris, which is a bit of a sore spot with me, as it would obviously be advantageous for the tech to know about OBP abd other Sparc/Solaris-unique issues. However, the FAA has systems that use other flavors of Unix (AIX and/or HP-UX) so it might not be ideal to use Sparc/Solaris alone, but I can't help but think that it would be better than using Redhat/PC. Anyway, perhaps I'll write a bit more about this and try to post it one day, to see what others here think...
Getting back on track, the perceptive reader will have already realized that the training of hundreds/thousands of military and FAA technicians in the ways of Unix will be good for the Unix community. I can use myself as an example: I've been working with the ASR-11 program for three years: my first contact with Unix was three years ago at the Raytheon ASR-11 school. Now I use a laptop running Redhat (previoulsy Solaris x86, but couldn't get the darn NIC to work), an Ultra 5 and a Sparc 5 on my desk/s at work, and at home I've two PCs with two drive-racks per PC, and swap between RedHat, Solaris x86, and XP. The FAA has also been generous enough to spring for a Sunkey memebership for me and I'm going to be doing as much training as possible this year. I'll have a go at the Sun SysAdmin tests later this year and then top it all of with the Sun Network Admin test. I'd then like to move into a part-time job with some local business that uses Sun boxes. I've discussed the possiblity with various classmates in the Sun courses I've taken, and apparently, and understandably, there's little demand for part-time network administrators. however, I'll settle for less: I'd really like the opportunity to hone my skills in the private sector, just to see how far my interest and talents might take me (perhaps out of civil sevice altogether and into the private sector full time? a fantasy perhaps, but one I occasionally indulge in).
To sum-up: the fallout from the implementation of these new systems will result in an even more widespread interest in Unix, and an enlarged geek contingent.