Computing & Air Traffic Control – Computerphile

Computing & Air Traffic Control – Computerphile


In the old days air traffic control was .. didn’t really exist as such. An airplane wanting to fly from one place to another would just take off and would try to work out how to get to wherever it was going. Over time it was realized that we needed some form of air traffic control in order to more or less to avoid airplanes from banging into one another, but also to provide them with a service so they actually would know what the conditions were enroute and so on and so forth. So during the fifties and sixties what evolved was the idea of an air traffic control center where once the airplane was away from the environment of the airfield we now had a control center which would then manage the aircraft through what’s called the Airways system. And the airways is up at high-level, it’s once you’ve got away from the airport then you’re into the airways system. The next problem was how do we actually make all that work. Now, in the old days you’d literally have had a controller in sitting in front of the radar and assistants running around with flight strips taking them from one controller to the other, and that clearly wasn’t going to work in the longer term. Essentially a flight strip, when a commercial flight flies it files what’s called a flight plan and the flight plan says what the the flight intends to do. Where its going from, where its going to. What its routing is, what type of aircraft it is and so on and so forth. Quite a a lot of information and essentially that ends up becoming a flight strip which follows that flight around. By the mid to late 60s we started to try to introduce computers into this, effectively to coordinate the movement of air traffic and particularly to ensure that the controller had the information that he required, but also that the hand over between controllers, as the flights progressed, was done in a structured kind of fashion. By the time we get down to nineteen seventy we actually have a control room that’s at west drayton and what that was doing was providing control for everything in the airways system from the south coast right the way up to the Scottish border. There’s an equivalent one at prestwick which does Scotland and then basically pushes the aircraft out across the Atlantic. Each one of these hoods here is effectively a radar sector an area of airspace which is controlled by that controller. So what you’ve got here is a couple of controllers, the director who is overall in charge of that sector and effectively what’s happening is as the flight progresses it moves from sector to sector until eventually disappears outside of the British air control controlled airspace. One of the early attempts was to put in a computer called the myriad, this is around the late sixties into about nineteen seventies. It takes a long time to develop this sort of software and by the time the software had actually been developed so reasonably, so it was in reasonably working condition air traffic had increased to such an extent that the myriad was no longer capable of doing the job. In fact it wasn’t capable of even doing the civil job. It was originally designed to do civil and military. It was designed for fault-tolerant environments so it was a triplex system. Three computer systems effective they’re all three computer systems were doing the same job and what would happen is that an arbitrator would decide who got the right answer. Hopefully all three got the same answer but if they didn’t then it would decide which was most likely answer and basically two out of three was the arrangement. We were continually chasing up against the limitations of what the computer systems could do. The next attempt was really when it all started working which was in 1972 or thereabouts– the Board of Trade as it was then decided that it was going to go out and procure the 53rd, i think it was, air traffic control systems from the United States because they already had this sectorised control system using big IBM mainframe computers. The IBM 360 is the underlying machine behind this. 10 million pounds of equipment in 1972 ish there abouts for the first time we now have enough processing power – more or less to be able to do the job completely. What’s being pictured here is some of the initial testing that’s being done before the system went operational. Testing all of the various different conditions and making sure that if we did nasty things to a bit over here a bit over there, they didn’t fall over. That’s actually me running with one of the tests sometime in 1974 it would be. I’d come back from the United States where I went to the training course at the end of 73, by 1974 i was in there I was a bit younger and thinner in those days. Sean>What did it feel like going over to the States and learning that, was that a new thing for you? It was an extraordinary thing, i was twenty barely twenty-two when I went out to states and and I was basically a happy technician. It was like all my christmasses come at once. As a youngster who was interested in technology to have a 10 million pound computer at my disposal was just just extraordinary. We were actually encouraged to, particularly when we were on the training course, we were encouraged to write software. So i decided one of the things that I decide that try and do is to write a game where the computer played Monopoly and you can see here the high-speed printers these printers would run at six hundred lines per minute. And i wrote this whole thing in assembler the computer would make a move and it would then print the Monopoly board. So this thing was throwing up pages like mad and as the game progressed so the computer had to do more and more work and so everything got slower and slower. So it wasn’t quite throwing the pages at the same sort of speed. Everything was paper-based. This was before the days of VDUs, we didn’t have VDUs and things like that. This is the system control here. The system control is effectively the center of gravity of this entire system. What we then have is three compute elements which were IBM 360-65s, which in their day were quite powerful computers. They had behind them, get this, three-and-a-half mega words of memory, core memory, three-and-a-half mega words. Yeah they’re 32-bit words. so what would that be around about forty megabytes or something of that order of RAM. And we ran the whole of the national air traffic service on that without any problem. In fact we could run it on about three of them. We had seven storage elements and we could run with three, we prefered to have four so each one was half a mega word. Then behind there was three input output controllers which were IBM 360-50s now the IOCE, input output control element, was responsible for talking to all the various different pieces of equipment that are attached to this. So we have flight strip printers, radar systems, all sorts of things all feeding information in or receiving information out from the 90-20 and they were managed by the input output control elements which you can’t see they’re just behind there. The system actually worked for quite a long time in fact it was in service as this system itself right the way through til 1990 when it was replaced by 4381 computers, a quarter / a third the size and something like 10 times more powerful but essentially running exactly the same software in emulation mode in “360 emulation mode.” So this was still at west drayton. And really that was mostly done because this was quite old equipment even when it was first put in and so it was becoming too expensive to maintain. The skills and knowledge necessary to maintain it were effectively becoming difficult to find now. So that was the reason that was done and that kept things going right the way through really until the early two thousands when it became apparent that really what was happening now was that computing was moving away from the large mainframe environment to much more distributed architecture. By now we have networking, we have the ability to have mini and micro computers which is where they where we want them rather than having in them in a major computer suite. So really the west drayton environment wasn’t designed for that, it was designed with the idea of having central processing bays and lots and lots and lots of cables running underneath ducts under the floor to where they’re going to be used. In addition the RAF who actually owned the west drayton site they kind of wanted it back so there’s lots of reasons why eventually in the early 2000s we started looking at the idea of moving to a new control center. This is swanwick and effectively that completely replaced the west drayton setup. Interestingly however, some of the code which was written way up here back in the nineteen seventies is actually still in use to this day down here. A friend of mine who worked together with me on a lot of this software, going back to the early seventies, he said “hey you know I was I was kicking around in some code the other day, just looking at it” and he says “and I came across your initials against a line of code,” he says, “I thought about you.” So the thing is if this software still works it ain’t broke so why fix it. The problem of air traffic control is continually evolving but the core functions are exactly the same. It takes a long time to get it all debugged, so we might as well keep with the code that we’ve already got. However over time eventually that code will be replaced but it’s it’s a slower process. This was a major exercise getting this to work in the in the 70s so they’ve really sweating the assets and taking advantage of the fact that its extremely reliable code. We’d like to thank Audible for supporting computerphile and if you go to audible.com/computerphile there’s a chance to sign up for a 30-day free trial and download a free audio book. The book I’d like to recommend is Ernest Cline’s Ready Player One. I enjoyed it so much I went off and downloaded his other book Armada as well, but ready player one is well worth checking out if you love the idea of a seamless VR system, dystopian future, lots of eighties pop culture references. I know a lot of people have been recommending ready player one. Its currently being made into a film so check it out and thanks once again to audible for supporting the computerphile channel. The other two canisters are actually butane cylinders which were discarded and they act kind of like a gas capacitor. They are called accumulators and they store gas close to where it comes out so that it can come out quickly

56 thoughts to “Computing & Air Traffic Control – Computerphile”

  1. I am learning to fly.
    Under VFR, it is still see and avoid (Unless you have flight following). It is pretty scary. I can't wait for ADS-B mandate to come into full compliance. It makes see and avoid a lot easier.
    Everything in the video is IFR.

  2. Wow! I really wouldn't have thought ATC moved away from mainframes until 2002! I'd have thought it would be much sooner.

  3. Interesting video. What also would be an interesting topic about computing in air traffic control is, how they maintain a failsafe computer network. Hardware always fails, but something like air traffic control has to compensate that in real time

  4. Impressive. Just shows you how well it was designed.

    "should we replace the system"
    "Nah just replace the hardware and run in 360 emulation"
    "we really do need to move away from mainframes now , show we re-write the code"
    " nah it will be fine if it aint broke don't touch it"

    It also fun that i am typing this comment on a keyboard that very well may have been connected to a IBM360 at some point in its life, controller replaced with a modern USB controller and she is still kicking after 36years of service, hears to another 30.

  5. It would be cool if you made videos of how companies use programming to actually make products and software. Obviously you have in the past with indie games and others. But you should actually show the code and how it's made just to help for more experiencew when us young programmers enter the field.

  6. I'm currently working on my thesis and in a part of it I use the higher level air traffic control architecture as an example. I enjoyed this video. Thanks, Sean and John.

  7. 1:16 Is it "Flight's Trip" or "Flight Strip"? Both sound exciting, with pilots and attendants and all, but y'know…

  8. Some what disparaging about the Myriad system which fulfilled the military role at West Drayton and did it in a stripless environment which the civil side is still struggling with at ATC centres.
    Signed Ex-Senior Engineer Myriad – 1980 til it was shut down

  9. "If it ain't broke why fix it" this is the kind of mentality that has held air traffic control system in the stone age since it started, the reality is, this system is far from appropriate for the volume of air planes today, this guy is very careful not to go into details on how it works, I wonder if they said "If it ain't broke why fix it" about the models of cars they had in the beginning of the 1900s, where we would be today.

  10. I wonder how many IBM 360s would "fit" into a Raspberry Pi Zero, either with emulation or with (optimized) recompilation for ARM.

    "if it ain't broke"… oh boy.

  11. "if it ain't broke, don't fix it" leads to stagnation and no improvements, which is antithetical to STEM as a whole. "if it ain't broke, don't fix it" Considered Harmful.

  12. By late-1970's we knew ATC was woefully 'under-puted'—PSA#182 was struck by a heart-attack/suicide-Cessna short of landing at crowded SAN; UAL#173 dry-tanked minutes shy of PDX because nobody, remembered, the FE tank panel had been reprogrammed; PATCO striked a few years later, and was fired; KAL007 flew near the USSR, and was shot down…

  13. Then the ATC union called a strike even though it was barred by law and Reagan fired all the ones who refused to report for work. Good times.

  14. In about 2005 I was working on an IBM assembler system that had been written in late 1960s for a large US company – it was written for 360 but I worked on S390 (Z systems I think they are called nowadays) – met an old colleague recently and she told me they finally switched off that assembler system last year …. so that ran commercially for way over 45 years

  15. So is the newest system running IBM 360 assembly code in some sort of emulation mode? Or how is code from 1974 still running on a more modern desktop machine?

  16. Yet another history video… could you please seriously consider going into modern computing problems in meaningful detail?

  17. I used to operate an IBM 360/135 mainframe computer. It had no VDU and programs were read into a card reader. It was then upgraded to a 360/138, which had a VDU and there was no need for the card reader. Those were such happy days.

  18. Figuratively speaking: Air Traffic Controllers are like semaphores, you need them in high density and traffic heavy locations but not on the highways.

  19. I worked on CM at Swanwick before it went live! My job was to keep software up to date on the various machines while development was ongoing. It was a lovely location—the café was in a large circular room that overlooked the lake. Beautiful!

Leave a Reply

Your email address will not be published. Required fields are marked *