Slashdot Mirror
FAA's Aging Flight-Plan System Having Problems
Eddytor takes us to eWeek for a look at the FAA's air-traffic control system, which, after 20 years of continuous operation, is in desperate need of an overhaul. Recent crashes have caused major delays, but the system's scope and importance make it difficult to test upgrades and improvements.
"Many technologies are used in air traffic control systems. Primary and secondary radar are used to enhance a controller's 'situational awareness' within his assigned airspace; all types of aircraft send back primary echoes of varying sizes to controllers' screens as radar energy is bounced off their skins. Transponder-equipped aircraft reply to secondary radar interrogations by giving an ID (Mode A), an altitude (Mode C) and/or a unique callsign (Mode S). Certain types of weather also may register on a radar screen."
I do wish TFS would make the distinction between software crashes and aircraft crashes.
Chernobyl 'not a wildlife haven' - BBC News
It doesn't need to make sense to me. If I handed a page of C++ to my grandmother, she couldn't make sense of that either. The weather report is concise and practical, giving a lot of information with the fewest amount of words. Once you can read it, you find it valuable to not have to sift through mounds of useless or redundant information (like adjectives, verbs, etc.)
Just because you can't read and understand it doesn't mean it doesn't have value to someone.
And what's that shit you posted at the end of your comment? Black People suck? Grow up, asshole.
, but the article doesn't give any real suggestions.
People probably won't like my suggestion, which would be to regulate air travel again. Cut the routes, limit take off and landing slots, increase the seat and isle widths and let airlines raise prices to the market level of support. Add a gas tax to keep the cost of gasoline above $3.50/gallon and take the money pay for building a high speed train system across the US. To me that would be worth going into debt for, short term anyway. It would create jobs here and give people an alternative to our broken air transportation system.
The trains could handle the commodity traffic and airlines could compete for luxury traffic, just like the old days. We have to do something. We have 3% of the world population and use 25% of the gasoline. Without alternatives we're never going to get people out of their cars. If I could go anywhere in the continental US in 24 hours, I'd never fly again.
With the added bonus of keeping air traffic at a predictable level for the FAA.
That's our life, the big wheel of shit. - The Fat Man, Blue Tango Salvage
It's pretty apparent that the current system isn't up to the task. I think the real questions should be more along the lines of upgrade or redesign? and in-house engineered versus contractor engineered? I hear there is a replacement on the way, but is it an actual 1 to 1 replacement or is it just replacing a few machines but the heart of the system is some old POS box that's been running since 1988? (I've seen other government networks receive upgrades like this)
Given the vast scale of the system, the constant use, and the time it would take to retrain all of the operators, how would you start testing and implementing new hardware? Just continue running the same code on new hardware... providing a few software tweaks to allow it to scale? Just how old is the current system? DOS era computing? CTOS? ENIAC?
greed@All_Evils:~#
TCAS isn't so much "in flight radar" as it is "holy shit last minute saver of your ass". TCAS doesn't do anything until a collision is basically imminent, at which point it gives instructions on how to avoid said collision.
ADS-B is the real in flight radar.
Name...That...Autocomplete!
I recently graduated from an aviation program at Purdue and I can tell you every single person I've ever sat down in a classroom with can read METARs, TAFs, and any other weather report just as quickly as if they were reading plain english.
More horseshit. I see cars on the side of the road almost daily on my commute. How often do you see a plane fall out of the sky because the engine died?
Name...That...Autocomplete!
The last time the FAA decided to do a major overhaul, they got a little too ambitious. They awarded a $4.5 billion contract to IBM to produce the Advanced Automation System, a complete replacement of the antiquated air traffic control system. The project was to begin with a major overhaul of the ATC workstations and human interface, looking at all the ideas engineers and air traffic controllers had to make the system better and safer. After 2 years IBM had blown through $2 billion and the only thing they had really accomplished was to replace the 1960s-vintage hardware with more recent gear. It was clear that it would take >$15 billion and >10 years to complete the project at the rate they were going, so the FAA cancelled the rest of the project. The less expensive $500 million version in Canada (CAATS, awarded to IBM's unsuccessful competitor Hughes Aircraft), was no more successful. Lesson learned: ATC system are *complicated*. They require near 100% reliability, and human lives depend on them. When they fail (as they must always do eventually), human controllers must be able to smoothly and safely pick up the entire workload in mid-flight, and then smoothly transition back to computer control when possible. Designing and implemnting this system is a challenge comparable to going to the moon.
I was an intern this summer at the FAA Technical Center. They are currently working on an overhaul of the national air space. The system that crashed a few weeks ago (the NADIN system) is in the middle of being replaced by NADIN II. They were testing it this summer. Also, look up the capstone program, its an effort to replace the radar based navigation with a GPS based system. ADS-B is a huge part of that, with the teams working on it winning the Collier award.
Once you can read it, you find it valuable to not have to sift through mounds of useless or redundant information (like adjectives, verbs, etc.)
You're suggesting that the your local TV station's Doppler 2008 15-minute weather segment is too long?
Dunno about you, but here in Southern California, getting the highs, lows, barometric readings, precipitation levels, wind speeds, wind directions, relevant surf, snow, rain or wind advisories, sun rise, sun set and current phase of the moon for where I live (and the same for a dozen or so nearby communities) from a friendly weatherman or weatherwoman that takes the time to describe and explain the relevance of all that information (hopefully with live footage, pictures, charts or graphs), is the only way to know with quantifiable certainty that tommorow's weather will be just like just like yesterday's and the day before that.
Unless, of course, you choose to look out the window or step outside long enough to realise you've probably got better things to do.
People, especially here in the US, are independent creatures. They prefer personal transportation to mass, and personal right now happens to be gas.
While people do often like their cars, as a person who has traveled by bus(both city and greyhound), train, plane, taxi, and car I have to say that there are reasons for so many people being almost glued to their vehicles.
To Wit: The alternatives suck. And the old saying: time costs money
For commutes, you're stuck using their schedule, not your schedule. When I had a *free* bus available, I mostly drove to work. Why? Because my work, despite being the one providing the bus, set the bus schedules in a paranoid fashion, resulting in adding 2 hours to my 12 hour work day. If it's simply added a half hour, I'd have taken it. The $2-4 saved back then just wasn't worth the time.
So, in any proposal to actually get people out of their cars, you have to acknowledge this. If you can make your theoretical public transport faster, cheaper, and more reliable than a car, you'd easily be able to get a large number of cars off the road.
That's why I like the idea of a high speed PRT system - you get the system's average speed above that of cars and a ticket that costs less than the gas to drive the same distance and you're gold. For an inner city system that'd often be 25-35 mph, for a interstate type system I'd want 100mph at a minimum*.
relax regulations on battery technology
Specifics?
*And a way to keep the same car when you stop to use the bathroom or even eat at a restaurant.
and allow more nuclear power plants
I agree with you here, but this reminded me of a local politician campaign add talking about 'adding more wind power to reduce our dependence on foreign oil'. I don't mind green energy by any means, but I detest fuzzy logic. Wind turbines produce electricity. Electricity, at this time, is insignificantly tied to our demand for oil. We could triple our electricity production and cut the cost in half and we'd barely reduce our demand for oil. At that, it'd be mostly people in the northeast switching from oil heating to electric. And they're already switching away from oil in many cases.
I don't read AC A human right
First let me say, I am no friend of the FAA. Everything in life is is a trade off between cost and risk. Everything. Period. In many cases, unless you're willing to pay $10,000 for your next coach plane ticket, many "fixes" are simply not cost effective given its low risk of occurrence.
Having said that, the FAA, as it relates to GA, is directly responsible for everything costing 2x or more than it should. They are also responsible for maintaining, if not elevating risk in many areas. Free market competition is vary rare for almost all aspects of GA aviation. Attorneys are directly responsible for all things GA aviation related costing a factor of 2 more than they should, in addition to the FAA's overhead.
If people really want to increase aviation safety, half the size of the FAA, require a pilot license to head the FAA, double the number of inspectors for commercial operators, and force a revamp of the certification process. As is, the FAA is directly responsible for keeping newer, safer, smaller, lighter technologies out of most cockpits and engine bays. Remember, it's a question of cost and everything aviation related is inflated 4x-8x higher than it would be if free market forces and liability protection would be allowed to function.
You are right about one thing, in many cases of aviation accidents, the FAA does have blood on its hands.
In more recent times, the spectre of the TSA has raised its head and is now starting to negatively impact aviation safety with no return on public safety. Does anyone remember the B2 bomber crash? Turns out some moister was the cause, inside some instrument pitot tubes. Now imagine TSA agents wilfully damaging the same types of instrumentation on commercial airliners in the name of public safety inspections; which are impossible to improve public safety. Recently, as many as 10 aircraft were ignorantly sabotaged by TSA inspectors in the name of public safety by climbing up onto the aircraft, on these very sensitive pitot tubes. Thankfully a pilot noticed some abnormalities and aborted his takeoff. Now keep in mind, it is impossible, regardless of the damage created, for these types of inspections to improve public safety.
Don't be fooled, the TSA is fighting hard to "get into the cockpit" and I have no doubt, public safety will continue to be compromised unless the public is educated on the dangers the TSA's well meaning yet ignorantly harmful involvement will cause. It's only a matter of time.
It it was only true!
There are so many points of failure in a system this complex, that it simply boggles the minds of the best architects we have out there.
Discloser... I am a pilot, I deal with Air Traffic Control and all the problems that they have
Let's begin with a single aircraft that will fly a from point A to point B. The flight is scheduled to leave at 0600Z from point A and arrive at point B at 1200Z for a total of 6 hours of flight time. The aircraft will have an SOA ( speed of advance ) of 600 kts ( nautical miles per hour ) and fly at 30,000 feet. Given this data the aircraft will cover 3600 nautical miles.
Given those parameters, it is simple to create and appropriate data structure that will represent the aircraft in question, allow us to create a series of data points to describe it's theoretical route, and predict where that aircraft is at any given moment with mathematical precision. In short it boils down to a rather simple database problem. Most any database cooker can come up with a set of queries to predict where crossing routes and position problems will be when you add more then one flight to the problem.
All of this will work just fine, right up until reality rears it's ugly head.
The cruise or en route portion of a flight is pretty much as simple as I have described, with the exception of having to readjust things based on headwinds, aircraft performance and other factors that may or may not change during the duration of the flight. We have gotten pretty good at predicting what the wind will be like at the planned altitude of the flight, but there are occasions when we are flat out wrong and have to make adjustments. If the winds at say 30,000 ft are not as predicted then to maintain the SOA the pilot needs to change altitude. So we can either propose a change, take that bit of data and run it through a "what if" calculation and then tell the pilot yes or no based on the result which will tell us if that action will cause a potential crossing problem with another flight, or have the software check all the flights currently in the system and have it give us an altitude that will not cause a crossing situation that is as close as possible to the desired altitude while maintaining a safety margin.
The real problem exists at the airports. Things get delayed, weather problems, mechanical problems, passenger problems, luggage problems, you name it, it is going to happen at one point or another. It backs the system up and then the simple database problem turns into the "Traveling Salesman Problem" from hell.
Let us consider a very probable occurrence..... Plane A is sitting at the gate getting serviced for the next flight. The fuel truck rolls up to full up the plane and the fueler gets out, gets his hoses out, plugs them into the fueling connection on the ground and connection on the plane. He looks at his manifest that reads 30,000 lbs of JET-A for this plane, he sets the controls on the fuel truck appropriately and starts pumping. For some reason when the meter reads 29,670 fuel starts spilling from the wing! His "Oh Fuck Light" goes of in his head and he runs for the truck to shut off fuel flow but by the time he makes it the 30 feet from where he is watching to make sure his connection is not leaking the meter now reads 29,980. So you have just spilled around 300 lbs ( about 50 gallons ) of fuel all through the wing and onto the ground. So this plane is not going ANYWHERE for at least the next couple of hours AT LEAST.
With this little problem, and it has happened to me things start to avalanche very quickly. I need another plane, another gate and I have to get the passengers and their luggage off of this plane, to the other plane at another gate, hint hint, this does not happen quickly. We are now occupying two gates and we are going to depart late, more then likely over an hour late if not a more.
So now the arriving flight that was supposed to park at the gate where the airpla
Hey KID! Yeah you, get the fuck off my lawn!
That's ridiculous, and a sign of complete stagnation on your part. How about we either fix the system, or design a better one? The answer is not to stagnate, but instead to build again!
Telling people to return to trains is ridiculous, and who has time for that anyway? If the air system isn't safe, fix it. If it can't be fixed, then build a better one. There is nothing that people in the 80's could do that we shouldn't be able to equal, if not vastly exceed. They weren't magicians, and their technology was far less advanced than what we have been able to create in the intervening two decades.
Where do I even start with this? Here are just a few of the many things wrong with this statement:
Beware of bugs in the above code; I have only proved it correct, not tried it.
You're only paying 2x as much? Sounds like a good deal. We had a Learjet a couple years ago that had a bad EL panel. Nothing fancy, just an EL panel with a handful of switches for radios. When the panel was previously replaced (about 4 years ago) it cost about $300. Two years ago, when we had to replace it again, it cost over $1200. A 4x markup in two years! Same P/N, and since there is no such thing as *new* for planes this old, it was also refurbished/remanufactured/rebuilt, however you want to look at it. I definitely blame the FAA and Insurance for the insane costs that we are seeing today.
"...TSA inspectors in the name of public safety by climbing up onto the aircraft, on these very sensitive pitot tubes."
Were these Rosemount styled pitot tubes by any chance? The costs for those things can be jaw-shattering. We had a lineman bend one on a jet right after it came back from RVSM installation and certification. He was trying to tug the plane with his car because the company's tug was broken down. He no longer works there.
I used to design air-traffic control systems.
The title and text of the parent post are inconsistent. The article is about the failures and obsolescence of the flight-plan system, but the discussion of radars, etc, in the text of the post is about other parts of the air-traffic control system. The flight-plan system interfaces to the part of the system that synthesizes radar data and allows communication from controllers to aircraft, but it is not that system. The reason for the interface is so you can do correlation of observed aircraft ID data, positions and position history with flight plans that have been filed. Then, if a plane goes off its flight path, the controllers can warn them and start emergency measures, which includes handing off to the air force.
The amount of data in a flight plan is pretty small, and the volume of messaging is on the order of a few million per year. Conceptually, NADIN is little more than a guaranteed-delivery email system. Next time they build the system they should consider routing over the Internet (of course using encryption) as a backup communication path. And there's also a huge amount that's been learned about system redundancy and scalability in the past few decades. The 99.9% uptime mentioned in the article is piss-poor for such a critical system. That's 8.76 hours per year of downtime. I delivered military systems in the 80's that had far better uptime. It wasn't even good in its own time.
I worked on both military and civilian air traffic control systems. The FAA and their consultants I met had that dangerous combination of arrogance and pig-ignorance that makes failure inevitable. They knew next to nothing about user interfaces, and had worse understanding of engineering tradeoffs than the average private sector middle manager (and that's pretty bad). By contrast, a good percentage of US Air Force officers involved in ATC actually knew what they were talking about. The FAA controllers I met were also shockingly ignorant of the capabilities and limitations of their systems, and some of their processes were there for historic reasons that no longer made sense. It was like dealing with overpaid DMV counter staff. It scares the hell out of me that people's lives depend on decisions that these knuckleheads make.
Get your teeth into a small slice: the cake of liberty
You're suggesting that the your local TV station's Doppler 2008 15-minute weather segment is too long?
IMHO, the short answer is "Yes."
I can get info for the next 48 hours on ONE page, with all the data I'll need. Don't believe me? Great! Let's try something [for those in the US.]
First, got to the NOAA's page. Enter your ZIP code in the upper left-hand side of the page.
Next, scroll to the bottom of the next page and click "Hourly Weather Graph" in the "Additional Forecasts and Information" section. Read the next page carefully. Try mousing over the graph for information on a particular data point.
That page has all the data I'll need to plan my days/weekend in one place. I can read it in less than 10 seconds. If I want radar/doppler, it's a link at the bottom of that page, and I can even get the doppler in motion, with a limited zoom function.
So yeah, even counting the time to pull up the page, enter a zip code, and click a link, it's my opinion that 15 minutes is too long to get the same info I can get in around a minute.
Don't tell me to get a life. I'm a gamer; I have LOTS of lives!