Science/Technology Question

[NorthSideSox72]

So, one of the deep sea vessels they are using to look through the wreckage of that Air France flight that crashed for unknown reasons in the deep Atlantic, recently found the flight recorder chassis, lifting hopes that they may find a usable memory unit to process.

 

Here is my question. Why, in the modern age, are these recorders still on the airplanes at all? Why not package up the data in chunks and send it in packets to collecting stations, to be stored on recursing hard drives for later use if needed? Its fairly simple data, and transmitters capable of bursting out that sort of data over thousands of miles are small and not that expensive. You wouldn't need a ton of data space either, because you could delete everything more than a few days old and just pull out what you need shortly after an event.

 

Anyone know why it isn't done that way? Just a cost thing?

 

[southsider2k5]

Knowing how things get with federal regulations, I would bet dollars to donuts that an old rule requires them to have a recorder on the plane so they can't legally do it. The the second problem would be that no one is set up for that sort of data storage, and doesn't want to incur the costs of it.

[NorthSideSox72]

Knowing how things get with federal regulations, I would bet dollars to donuts that an old rule requires them to have a recorder on the plane so they can't legally do it. The the second problem would be that no one is set up for that sort of data storage, and doesn't want to incur the costs of it.

I think the data storage is the easy and cheap part. Its not a huge ton of data. The hard part, I think, is partially the regulations you mentioned, and partially the cost the airlines would incur in changing their aircraft to add on the transmitters.

 

[southsider2k5]

I think the data storage is the easy and cheap part. Its not a huge ton of data. The hard part, I think, is partially the regulations you mentioned, and partially the cost the airlines would incur in changing their aircraft to add on the transmitters.

 

That is a lot of data though. Every single bit of info from every single plane trip? Maybe if they figured out a way to not store it for very long it could work. But I have a feeling it would involve a significant amount of archiving if they made those changes.

[StrangeSox]

That is a lot of data though. Every single bit of info from every single plane trip? Maybe if they figured out a way to not store it for very long it could work. But I have a feeling it would involve a significant amount of archiving if they made those changes.

 

If they're logging it on the plane, I don't see why the storage would be more cumbersome or expensive on the ground. Currently they record something 24 hour chunks and re-write over old data on the planes.

 

According to wikipedia (supreme authority), the space shuttles used this sort of data logging and didn't have onboard FDR's. I'd imagine that's due to the likelihood of an FDR burning up or being lost in space.

 

Here's the FAA's TSO on FDR's, FYI:

 

http://www.airweb.faa.gov/Regulatory_and_G...;FILE/C124a.pdf

[NorthSideSox72]

That is a lot of data though. Every single bit of info from every single plane trip? Maybe if they figured out a way to not store it for very long it could work. But I have a feeling it would involve a significant amount of archiving if they made those changes.

Well for one, the data is quite simple - its just a series of states that could be represented in short data fields, repeated continuously. One plane might generate a thousand bytes of data, tops, in each burst. If the bursts occur on the second, that's 60k per minute, 360k per hour, say an average flight time of 3 hours globally makes it about a megabyte of data per flight. In a day globally, there are what, 10k flights? That's 10,000 MB, which is 10 GB. If you store 7 days' worth, that's 70 GB. That's a small server, nothing huge.

 

If an event occurs where data is needed, and the data is keyed with airline/flight info, then FAA/NTSB/ESA or whomever can just pull that data off for analysis. The data can then recusively be wiped after the 7 day window.

 

And even if you wanted to keep it for a year, since its simple data, you just zip it down and put it on a tape somewhere.

 

Storage is the least expensive part of this proposition, along with maintenance which could be just a few human beings at most. The expensive parts are regulatory changes and agreements, and the wildcard - I have no idea what the cost of the transmitters would be per aircraft.

[StrangeSox]

They're storing it on individual drives on each plane, no reason they couldn't have the same storage capacity but substantially cheaper on the ground. Disk space is incredibly cheap these days.

[southsider2k5]

If they're logging it on the plane, I don't see why the storage would be more cumbersome or expensive on the ground. Currently they record something 24 hour chunks and re-write over old data on the planes.

According to wikipedia (supreme authority), the space shuttles used this sort of data logging and didn't have onboard FDR's. I'd imagine that's due to the likelihood of an FDR burning up or being lost in space.

 

Here's the FAA's TSO on FDR's, FYI:

 

http://www.airweb.faa.gov/Regulatory_and_G...;FILE/C124a.pdf

 

That's what I think would change. I bet if they took it off of the planes, they would do a lot more archiving.

[Milkman delivers]

If they're logging it on the plane, I don't see why the storage would be more cumbersome or expensive on the ground. Currently they record something 24 hour chunks and re-write over old data on the planes.

 

According to wikipedia (supreme authority), the space shuttles used this sort of data logging and didn't have onboard FDR's. I'd imagine that's due to the likelihood of an FDR burning up or being lost in space.

 

Here's the FAA's TSO on FDR's, FYI:

 

http://www.airweb.faa.gov/Regulatory_and_G...;FILE/C124a.pdf

 

Trying to be confusing or just in a very serious rush?

 

When was this crash, anyway? I don't believe I've heard about it.

[StrangeSox]

Trying to be confusing or just in a very serious rush?

 

Had to throw in the extra FYI!

 

When was this crash, anyway? I don't believe I've heard about it.

 

2009. NOVA had a good special on possible causes of the crash.

[LittleHurt05]

When was this crash, anyway? I don't believe I've heard about it.

 

June 1, 2009

Edited April 28, 2011 by LittleHurt05

[southsider2k5]

That's what I think would change. I bet if they took it off of the planes, they would do a lot more archiving.

 

I bet they could use it to track other flight disasters. Thinking back to the Southwest Airlines flight that recently went convertable, I wonder if they could gone back and caught something in flight pressures to show that the plane was cracking for example.

[StrangeSox]

I bet they could use it to track other flight disasters. Thinking back to the Southwest Airlines flight that recently went convertable, I wonder if they could gone back and caught something in flight pressures to show that the plane was cracking for example.

 

That sort of failure was likely brittle failure and catastrophic, meaning the crack would have formed, grown for a while but not actually breached the hull, and then failed rapidly.

[southsideirish71]

From a technological standpoint this isnt a huge stretch. Store and forward has been out for a while( forever ). From a data size standpoint its the voice recorder traffic that would have the most impact. The instrumentation data shouldn't be in a format that requires lots of data or packet size. The key issue is around integrity. You can encrypt this and use a satellite connection to spit this out with a second or two delay. You would just need some process to either hash the data for forensic integrity. IPSec has some of this built in, however you want to forensically mark and timestamp records. If integrity wasn't a huge issue you could use a dedup process like a wan/satellite optimization device to keep the uplink clean. With repeatable data, only send changes. In reality from a bandwidth standpoint, this isnt that much of an issue. You see in flight wifi now being offered. The overhead shouldnt be that much more than a few business travellers poking around on the internet.

Edited April 28, 2011 by southsideirish71

[NorthSideSox72]

From a technological standpoint this isnt a huge stretch. Store and forward has been out for a while. From a data size standpoint its the voice recorder traffic that would have the most impact. The instrumentation data shouldn't be in a format that requires lots of data or packet size. The key issue is around integrity. You can encrypt this and use a satellite connection to spit this out with a second or two delay. You would just need some process to either hash the data for forensic integrity. IPSec has some of this built in, however you want to forensically mark and timestamp records. If integrity wasn't a huge issue you could use a dedup process like a wan/satellite optimization device to keep the uplink clean. With repeatable data, only send changes. In reality from a bandwidth standpoint, this isnt that much of an issue. You see in flight wifi now being offered. The overhead shouldnt be that much more than a few business travellers poking around on the internet.

You make a good point about voice, that's data that is more dense and takes up more space than the simple diagnostic data. Still though, the overall data space is small in terms of server world. And as you also say, with the diagnostic data, it can be dynamic to only send updated values in some sort of tag-value pairs to cut down data width from mostly-repeated fixed format lines. This should be very doable.

 

[Milkman delivers]

Had to throw in the extra FYI!

 

 

 

2009. NOVA had a good special on possible causes of the crash.

 

Thank you.

 

I know this isn't the thread for it, but I like the theory of methane gas being released from undersea faults and causing fires/explosions on ships and planes to explain the disappearances in the Bermuda Triangle. I remember seeing a special on it and one scientist proposed that theory. It seemed the most believable.

[maggsmaggs]

Here's a question. How the hell do you invent the Internet? And how the f*** does it actually work? My mind is boggled by this question daily.

[Balta1701]

Here's a question. How the hell do you invent the Internet? And how the f*** does it actually work? My mind is boggled by this question daily.

If we link computers together and they back up each other, then if the Russkies nuke us, we won't lose command and control, so we can nuke them back.

[Leonard Zelig]

Had to throw in the extra FYI!

Gotta love the TLAs.

[southsider2k5]

Thank you.

 

I know this isn't the thread for it, but I like the theory of methane gas being released from undersea faults and causing fires/explosions on ships and planes to explain the disappearances in the Bermuda Triangle. I remember seeing a special on it and one scientist proposed that theory. It seemed the most believable.

 

Or so the government would have you believe...

 

No but seriously I saw that one too. It was very compelling, and interesting when they did the tests on the model ship.

[southsider2k5]

Here's a question. How the hell do you invent the Internet? And how the f*** does it actually work? My mind is boggled by this question daily.

 

Ask Al Gore.

[Milkman delivers]

Or so the government would have you believe...

 

No but seriously I saw that one too. It was very compelling, and interesting when they did the tests on the model ship.

 

The best demonstration I've seen of the effect was actually on Jay Leno like 10 years ago. Some guy had an aquarium filled with water and a small nozzle at the bottom that released the gas. I know it showed the planes bursting into flames. I can't recall if the boats did the same, or if they just lost buoyancy and sank immediately.