Black Boxes (FDRs)
Thread Starter
Join Date: Aug 2006
Location: Nairn, Highland
Age: 85
Posts: 159
Likes: 0
Received 0 Likes
on
0 Posts
Black Boxes (FDRs)
I suspect that I am about to re-invent the wheel and have doubtless missed earlier discussions. However....
Recent events have indicated that at least some in-flight data is already being fed to the ground. It cannot be beyond the bounds of modern technology to feed (perhaps via satellite) ALL the information that is currently stored on the Flight Data Recorder/Cockpit Voice Recorder thereby obviating the need to physically recover the Black Box in the event of an accident.
This surely is not a new idea?
I appreciate that many safeguards would have to be built in to any new system. e.g. the data would only be available AFTER an event and not, for example, to allow Management to eavesdrop on routine flight deck conversations.
Jack
Recent events have indicated that at least some in-flight data is already being fed to the ground. It cannot be beyond the bounds of modern technology to feed (perhaps via satellite) ALL the information that is currently stored on the Flight Data Recorder/Cockpit Voice Recorder thereby obviating the need to physically recover the Black Box in the event of an accident.
This surely is not a new idea?
I appreciate that many safeguards would have to be built in to any new system. e.g. the data would only be available AFTER an event and not, for example, to allow Management to eavesdrop on routine flight deck conversations.
Jack
With all the parameters that modern FDR's record the amount of data that would need to be stored would be mind boggling. The other issue is confidentiality so the data would have to be encrypted. Its not impossible but its not as straightforward as it would seem.
Join Date: May 2001
Location: England
Posts: 1,904
Likes: 0
Received 0 Likes
on
0 Posts
I can assure you that in modern computing terms the amount of data a FDR captures is tiny. Sure there might be a couple of hundred different parameters but we are talking about logging parameters (yes/no/on/off) and values (time, position, deflection, alt, speed, aoa etc), not amazing detail digital graphics. This data is then read, replayed and analysed by software on the ground. The data is logged as simple text. In the amount of time it took to load this page on pprune (a page which included graphics and special formatting) we could've downloaded a minutes worth of FDR data at least and that's without compression. Let's not forget that two way satellite based telephone communications have been possible for over 10 years and Internet access has also now been possible for a number of years. The amount of bandwidth needed to transport voice data and internet traffic is far greater than that of what an FDR needs.
For comparision, this from Wikipedia: "According to the publisher of the OED, it would take a single person 120 years to type the 59 million words of the OED second edition, 60 years to proofread it, and 540 megabytes to store it electronically"
Therefore:
59,000,000 words = 540Mb
590,000 words = 5.4Mb
590 words = 0.054Mb or approx 54Kb
I can't imagine a FDR churning out more than a couple of hundred average length words per second. Even if it did we're talking about needing the ability to send this data at approx 50Kbps, i.e. last century speeds. Securing the data is a given. The larger airlines have departments responsible for downloading the data and analysing it. Therefore apart from implementing this solution the only issue for airlines is that of a tiny bit of training and the upkeep of servers that can receive this information around the clock.
I can see how in the past there would have been an extreme privacy issue around FDR data but this should be revisited by ICAO and the manufacturers.
For comparision, this from Wikipedia: "According to the publisher of the OED, it would take a single person 120 years to type the 59 million words of the OED second edition, 60 years to proofread it, and 540 megabytes to store it electronically"
Therefore:
59,000,000 words = 540Mb
590,000 words = 5.4Mb
590 words = 0.054Mb or approx 54Kb
I can't imagine a FDR churning out more than a couple of hundred average length words per second. Even if it did we're talking about needing the ability to send this data at approx 50Kbps, i.e. last century speeds. Securing the data is a given. The larger airlines have departments responsible for downloading the data and analysing it. Therefore apart from implementing this solution the only issue for airlines is that of a tiny bit of training and the upkeep of servers that can receive this information around the clock.
I can see how in the past there would have been an extreme privacy issue around FDR data but this should be revisited by ICAO and the manufacturers.
Registered User **
Join Date: Mar 2009
Location: USA
Age: 49
Posts: 480
Likes: 0
Received 0 Likes
on
0 Posts
A full dump is about 35 meg. Transmitting realtime is not alot of data at all. Satcom is very secure and the data is not really much of a security issue.
This in my oppinion is on the way as hardware like gatelink is becoming more common. The more computers we put in our aircraft the more data there is for us to update the computers. Satcom and irridum transmissions are solid and will probably be used for updating in the future.
The other thing to add is FANS, the motivation for FANS is supposed to be to increase air/ground communication accuracy, especially as aircraft overfly forign territory to reduce the safety issue of miscommunication. In all actuality the real Future Air Navigation Systems will be tied into FMS allowing ATC to command and control an aircraft. Pilots are slowly being replaced by technology.
Back on topic, constant data transmission at this point in time is not cost effective but absolutely possible. It would be great to have it set up where an ACARS systems message triggers an acms data stream say for 5 minutes then cancels if all parameters seem normal and further failures do not exist.
This in my oppinion is on the way as hardware like gatelink is becoming more common. The more computers we put in our aircraft the more data there is for us to update the computers. Satcom and irridum transmissions are solid and will probably be used for updating in the future.
The other thing to add is FANS, the motivation for FANS is supposed to be to increase air/ground communication accuracy, especially as aircraft overfly forign territory to reduce the safety issue of miscommunication. In all actuality the real Future Air Navigation Systems will be tied into FMS allowing ATC to command and control an aircraft. Pilots are slowly being replaced by technology.
Back on topic, constant data transmission at this point in time is not cost effective but absolutely possible. It would be great to have it set up where an ACARS systems message triggers an acms data stream say for 5 minutes then cancels if all parameters seem normal and further failures do not exist.
In all actuality the real Future Air Navigation Systems will be tied into FMS allowing ATC to command and control an aircraft.
I doubt that ATC will ever have the authority to command and control the aircraft.
I can see how in the past there would have been an extreme privacy issue around FDR data but this should be revisited by ICAO and the manufacturers
Not sure why the extreme privacy issues have changed and in fact with the trend to litigation instead of safety enhnacemen, its even more reason not to allow FDR and CVR data to become less secure.
Maybe the solution is to use GPS tracking capabilities to make it easier to find the CVR and FDR. This would also be of assistance in locating the wreckage. If the latest ELTs have that function why can't it be built into the "black boxes"?
I doubt that ATC will ever have the authority to command and control the aircraft.
I can see how in the past there would have been an extreme privacy issue around FDR data but this should be revisited by ICAO and the manufacturers
Not sure why the extreme privacy issues have changed and in fact with the trend to litigation instead of safety enhnacemen, its even more reason not to allow FDR and CVR data to become less secure.
Maybe the solution is to use GPS tracking capabilities to make it easier to find the CVR and FDR. This would also be of assistance in locating the wreckage. If the latest ELTs have that function why can't it be built into the "black boxes"?
Join Date: Apr 2008
Location: November18
Age: 48
Posts: 258
Likes: 0
Received 0 Likes
on
0 Posts
Hf data-link is more than capable of being encrypted and transmitting a whole host of parameters, as well as radar positioning. I see a problem with the "net" having to be constantly managed though due to the need for freq changes etc.
Anybody here familiar with "Link 16"? As used in the RN.
Anybody here familiar with "Link 16"? As used in the RN.
Join Date: Apr 2009
Location: Melbourne, Australia
Age: 38
Posts: 42
Likes: 0
Received 0 Likes
on
0 Posts
Even if the data was being transmitted, in my opinion you would need to keep the FDR in the aircraft, as it would be very difficult to guarantee the integrity of the data transmission when the aircraft is in an emergency situation, i.e. during a departure from controlled flight in a roll or spin.
Join Date: Jul 2007
Location: Germany
Posts: 1
Likes: 0
Received 0 Likes
on
0 Posts
HF ALE overcomes this issue. You upload a list of frequencies, the box automatically finds the best connection to use.
Problem is, the data rates are SLOOOW and you cannot guarantee always getting a connection due to atmospherics.
Satcom is definately the way ahead, most of us have APDLC and such like, I cannot think however how many more sats we would need to support ALL aircraft doing this...
Satcom is surprisingly slow as a data transmission method due to bandwidth issues with all those users on the net.
Cost, I suggest is the major issue.
Other factor, how many accidents lead to an unrecoverable FDR? In recent times, I can think of only one. I dont wish to sound cruel, but whatever has happened to the AF flight is tragic, but aviation is inherently not 100% safe... I honestly dont think the number of FDR losses could ever justify launching dozens of more satelites.If we all have to re-learn not to fly into big thunderstorms, take lots of extra fuel in the tropics and realise sometimes you need to turn around rather than carry on to destination, surely we can cover 90% of the scenarios that are likely to have brought down the AF flight? Sometimes that 90% is the best we can do.
One other option would be for data replication however, whereby a single UHF radio could beam FDR data to other aircraft in UHF range as a budy-budy data backup system. Perhaps not infalliabe, but certainly more viable than Satcom relay of FDR feeds.
Problem is, the data rates are SLOOOW and you cannot guarantee always getting a connection due to atmospherics.
Satcom is definately the way ahead, most of us have APDLC and such like, I cannot think however how many more sats we would need to support ALL aircraft doing this...
Satcom is surprisingly slow as a data transmission method due to bandwidth issues with all those users on the net.
Cost, I suggest is the major issue.
Other factor, how many accidents lead to an unrecoverable FDR? In recent times, I can think of only one. I dont wish to sound cruel, but whatever has happened to the AF flight is tragic, but aviation is inherently not 100% safe... I honestly dont think the number of FDR losses could ever justify launching dozens of more satelites.If we all have to re-learn not to fly into big thunderstorms, take lots of extra fuel in the tropics and realise sometimes you need to turn around rather than carry on to destination, surely we can cover 90% of the scenarios that are likely to have brought down the AF flight? Sometimes that 90% is the best we can do.
One other option would be for data replication however, whereby a single UHF radio could beam FDR data to other aircraft in UHF range as a budy-budy data backup system. Perhaps not infalliabe, but certainly more viable than Satcom relay of FDR feeds.
Guest
Posts: n/a
Maybe a dumb question as I haven't seen it asked by anyone else but is it possible to make FDRs floatable? I know they're pretty dense but there has got to be a way to make them buoyant, eg by using a polyurethane foam casing or some other ultra low density material.
It's such a shame that they're otherwise pretty indestructable but that's of absolutely no use when they're functioning perfectly where no one can find them, under 20,000 feet of water.
It's such a shame that they're otherwise pretty indestructable but that's of absolutely no use when they're functioning perfectly where no one can find them, under 20,000 feet of water.
Although I have been influenced by the Air France crash to ask this, i'm not speculating as to the cause, but if an aircraft was suffering from air data faults or icing, would the FDR provide false information? Would it just tell the story as the pilots could see it, or would it have separate inputs as well as recording those from the flight deck?
Join Date: Sep 2002
Location: La Belle Province
Posts: 2,179
Likes: 0
Received 0 Likes
on
0 Posts
The FDR is dependent upon the same sources as the flight crew. So if the air data system is disabled then the same inaccurate data will be recorded by the FDR.
The only circumstances where the FDR can record data different to the crew is when the source data is ok, but the display system is faulty. For an extreme example, if all the cockpit displays were unservicable, the FDR would record the correct info, but the crew would be unable to see any of the data.
Even if the FDR is recording the same (unreliable) data that is of value because
(a) it indicates that the soruce of the data was the problem; and
(b) with the luxury of time and a calm ground environment, it can be possible to deduce reality even from multiple sets of misleading data - because we know the aircraft conforms to basic physical laws, you can cross-compare data and synthesise, say, a flight path, then check against other data. The hypothesis which best fits the known facts can then be assumed valid, and thus "right" can be sorted from "wrong".
The only circumstances where the FDR can record data different to the crew is when the source data is ok, but the display system is faulty. For an extreme example, if all the cockpit displays were unservicable, the FDR would record the correct info, but the crew would be unable to see any of the data.
Even if the FDR is recording the same (unreliable) data that is of value because
(a) it indicates that the soruce of the data was the problem; and
(b) with the luxury of time and a calm ground environment, it can be possible to deduce reality even from multiple sets of misleading data - because we know the aircraft conforms to basic physical laws, you can cross-compare data and synthesise, say, a flight path, then check against other data. The hypothesis which best fits the known facts can then be assumed valid, and thus "right" can be sorted from "wrong".
Join Date: Jul 2007
Location: Germany
Posts: 1
Likes: 0
Received 0 Likes
on
0 Posts
Would imagine GS output would be one of the DataWords recorded.
Compared to the timestamp, a delta distance/Time can be calced, thus a groundspeed.
Comparing known local winds with Calibrated recorded airspeed, any major discrepancy would show up any incorrect data feed.
Compared to the timestamp, a delta distance/Time can be calced, thus a groundspeed.
Comparing known local winds with Calibrated recorded airspeed, any major discrepancy would show up any incorrect data feed.
Join Date: Sep 2002
Location: La Belle Province
Posts: 2,179
Likes: 0
Received 0 Likes
on
0 Posts
I'd be reluctant to use forecast wind info to try to ascertain which of several disagreeing airspeeds (if any) might be correct. (I doubt there's such a thing as "known" local winds accurately on the flight path)
You'd tend to try to reconstruct the aircraft flight path and work out the discrepancies.
If I know the aircraft configuration (cruise makes that easy in this case) and have normal acceleration recorded (nominally at 8Hz) and angle of attack recorded (at maybe 2 Hz, I cant recall what the reg calls for) then as OEm I can workout what the speed must have been, since only one speed will give that 'g' for that AoA. Do that for every data point on a reconstructed flight path, and you'll have a reconstructed airspeed you can now compare to the recorded ones. Hopefully they start in agreement somewhere (or all have a common offset, so you can correct any errors in the config data, such as weight). You then move foreward and look for the divergences from the constructed data, and try to build a hypothesis for why its going on.
You'd tend to try to reconstruct the aircraft flight path and work out the discrepancies.
If I know the aircraft configuration (cruise makes that easy in this case) and have normal acceleration recorded (nominally at 8Hz) and angle of attack recorded (at maybe 2 Hz, I cant recall what the reg calls for) then as OEm I can workout what the speed must have been, since only one speed will give that 'g' for that AoA. Do that for every data point on a reconstructed flight path, and you'll have a reconstructed airspeed you can now compare to the recorded ones. Hopefully they start in agreement somewhere (or all have a common offset, so you can correct any errors in the config data, such as weight). You then move foreward and look for the divergences from the constructed data, and try to build a hypothesis for why its going on.
Join Date: Jan 2008
Location: LONDON
Age: 51
Posts: 525
Likes: 0
Received 0 Likes
on
0 Posts
Using sat comms is most definately possible and to return to superpilots post when transmitting information you would use mnemonics rather than words - so you mark up the data into small mnemonics so you transmit only bare information (which also can be easily compressed).
The bandwidth and the transmission are still the problems, whilst a single aircraft is not an issue if you look at the number of flights across the world at any given point you start to get a scale of the costs involved (if I recall correctly from a conversation a while ago it was around £5k per month to rent a single 64k channel on Inmarsat and that was for only a single satellite) - also as pointed out here previously such information can only reliably be transmitted when in stable flight because of alignment issues between the atennae and the satellite.
VinRouge hit on an interesting point in replicating to nearby aircraft - this is more practical due to the use of omnidirectional antennas for HF/VHF etc but it does pose some issues which can be overcome but require a global change across all fleets.
The problems are mainly that seperate portions of a FDR log could be stored across many different aircraft (i.e. an aircraft being replicated to may be travelling in the opposite direction so will be only in range for a shortish period) - these would need to be gathered together if required which would involve a logistical problem in identifying which aircraft had which data (not insurmountable). This also raises other issues such as if you have multiple aircraft in the airspace who replicates to whom and how long are they going to be required to keep it - this puts a dependency on each aircrafts FDR having enough capacitiy to store not only its own information but that of others - technology is there already to cope with this but you require all flights to be technology capable to that extent required.
As has already been pointed out the number of incidents where the FDR is not recovered is minimal to a point that the huge cost and logistics of retrofitting such a technology is prohibitive - it may be a directive one day that any new purchases work along similar lines to the above but although its out of my field I personally cannot see it being viable to retrofit.
The bandwidth and the transmission are still the problems, whilst a single aircraft is not an issue if you look at the number of flights across the world at any given point you start to get a scale of the costs involved (if I recall correctly from a conversation a while ago it was around £5k per month to rent a single 64k channel on Inmarsat and that was for only a single satellite) - also as pointed out here previously such information can only reliably be transmitted when in stable flight because of alignment issues between the atennae and the satellite.
VinRouge hit on an interesting point in replicating to nearby aircraft - this is more practical due to the use of omnidirectional antennas for HF/VHF etc but it does pose some issues which can be overcome but require a global change across all fleets.
The problems are mainly that seperate portions of a FDR log could be stored across many different aircraft (i.e. an aircraft being replicated to may be travelling in the opposite direction so will be only in range for a shortish period) - these would need to be gathered together if required which would involve a logistical problem in identifying which aircraft had which data (not insurmountable). This also raises other issues such as if you have multiple aircraft in the airspace who replicates to whom and how long are they going to be required to keep it - this puts a dependency on each aircrafts FDR having enough capacitiy to store not only its own information but that of others - technology is there already to cope with this but you require all flights to be technology capable to that extent required.
As has already been pointed out the number of incidents where the FDR is not recovered is minimal to a point that the huge cost and logistics of retrofitting such a technology is prohibitive - it may be a directive one day that any new purchases work along similar lines to the above but although its out of my field I personally cannot see it being viable to retrofit.
Join Date: Aug 2005
Location: fairly close to the colonial capitol
Age: 55
Posts: 1,693
Likes: 0
Received 0 Likes
on
0 Posts
The trouble with streaming DFDR data is cost. As already mentioned, it is a rarity when the boxes are not found. On the other hand, SATCOM data links charge by the stream volume, usually in tiers, and per connection point.
100 aircraft in a smallish-medium sized carrier's fleet streaming constant data would add up to a considerable sum very quickly.
Something that should be considered is, as moribund as it might be, pre-ditch pings of location data. Other circumstances like multiple system failures, high-G events and other serious situations where a happy outcome might be in question, even if 'minorly' so. Better to be safe than sorry in such cases.
100 aircraft in a smallish-medium sized carrier's fleet streaming constant data would add up to a considerable sum very quickly.
Something that should be considered is, as moribund as it might be, pre-ditch pings of location data. Other circumstances like multiple system failures, high-G events and other serious situations where a happy outcome might be in question, even if 'minorly' so. Better to be safe than sorry in such cases.
Registered User **
Join Date: Mar 2009
Location: USA
Age: 49
Posts: 480
Likes: 0
Received 0 Likes
on
0 Posts
vapilot2004
A possible solution could be a functional command to initate data transmissions when the aircraft encounters a distressed condition. Possibly initiated by crew when they understand (new sop's created) they have encountered a dangerous environment or systems failure as a backup.