"One thing we can all agree on in aviation accidents is there's never one single cause of an accident. It's a long chain of events that cause an accident."
This certainly rings true for me, both within and outside aviation. I've had an interest in "big" accidents and how they might be prevented for decades. For the aviation industry especially, it's been remarkable how few accidents and deaths there have been in the last couple of decades, versus previous decades when such accidents and loss of life were in the news several times a year.
I'm not sure what first sparked my interest in this, but I recall the TV news coverage the day (probably within an hour or two of the event, I was fascinated with the accident coverage and was late getting to work) the Space Shuttle Challenger exploded after liftoff in 1993. One of the reporters was saying that in recent years he had spoken to an astronaut about the shuttle program, and he quoted the astronaut "you know, someday one of these things is going to blow up." The reporter further interpreted the astronaut's statement, saying "he had a PREMONITION that this would happen."
I wanted to throttle that reporter. The reporter seemed to imply that the astronaut had some "special" or metaphysical knowledge.
It had always been obvious to me as an engineer that the Shuttle was an exceedingly complex thing, that while every effort was made to make every launch a success, there's no possible way engineering-wise to guarantee that it would work right 100 percent of the time, and the astronaut was just stating the risk they all knew they were taking with every flight.
In the late 1990s I had a subscription to Embedded Systems Journal. One article of interest was on the Therac 25, a medical instrument designed to give precise amounts of radiation for cancer treatment. Every once in a while, it gave, by far, too much radiation. It might be easy to say the whole blame for this was software, but (IMHO) there also could have been hardware protections against it happening, but apparently the designers thought things would work just fine as-is, and the thought of having any backup or "belt-and-suspenders" protection may not have even crossed their minds. The device caused several injuries and deaths, perpetuated by engineers at the manufacturer denying any problem. Wikipedia is just one of many sources of info on what happened:
Therac-25 - Wikipedia
Another blowup, this one fairly easily blamable on software (and lack of procedures for good testing, etc.) is the first flight of the Arian 5 rocket.
ARIANE 5 Failure - Full Report
Then there was the Shuttle Columbia burning up on re-entry in 2003. The story was chronicled in a magazine article that was reprinted in the book "The Best American Science Writing 2004" where I read it. Like the Challenger accident, there were many things that went wrong, yet in both cases engineers in charge (unlike the Therac 25 case) saw a potential problem (in the Challenger, Mortin Thiokol engineers before launch saying it was too cold, and Columbia while in orbit, those who watched launch videos and saw something significant hit a wing) and tried to do the right thing.
Related to complex systems and their failure, here are two things that are NOT recent, but that I only recently came across:
Swiss cheese model - Wikipedia
http://web.mit.edu/2.75/resources/random/How Complex Systems Fail.pdf
This book is a very meaty treatment of the issues surrounding the Challenger, I'm part-way through but highly recommended : https://www.amazon.com/Challenger-L...lenger+disaster&qid=1569811352&s=books&sr=1-2
The real reason is that a single phone being used on a plane can wreak havoc on the cellular system, a bunch can cause a bottleneck or local system overload and shutdown.
In normal operation a cell phone will hit several towers at once. The system must figure out which tower will talk back to the phone, then track it as it moves through the system and seamlessly transfer the call from tower to tower.
Each phone carrier (AT&T, Verizon....) has a team of people optimizing this process daily based on recent tower data. One of my friends became the Verizon "can you hear me now" guy after he was laid off at Motorola where we both worked. His job was to drive one of several routes daily with a computer attached to a phone logging data.
One phone at an altitude of a few thousand to tens of thousands of feet can hit a lot of towers causing the system to react in an unpredictable manner. At low altitudes it can also move faster than the system can track it.
A phone inside a metal plane will have an extremely odd RF radiation pattern due to the multiple windows, so it's possible to hit towers in two different geographical locations that are on totally different sub systems, therefore starting two different calls at once.....the cell systems really don't like this.
In recent years the systems have employed more and more towers with ever decreasing range to improve system capacity. The antennas are aimed downward in many areas for this reason. The down tilt does reduce interference from phones in the sky, but it can still mess up the system.
Note, that I left the cell phone design world over 10 years ago, and things have changed a lot, but the basic principles of operation have not changed.
But after take off and during flight operation is allowed. Is that different then?
Jan
back in the analog days being in a plane would trigger the anti-cloning on the network as you couldn't be in two places at once. As GSM tells the phone which cell to talk to there is less of an issue. WCDMA everything uses the same frequency anyway.
But if you look at vertical beamwidth on cell tower antennas, not much goes up, so you'll lose coverage quickly and the networks can't go more than a few dozen km anyway. You then have basestations on the plane itself for in flight calling etc.
But if you look at vertical beamwidth on cell tower antennas, not much goes up, so you'll lose coverage quickly and the networks can't go more than a few dozen km anyway. You then have basestations on the plane itself for in flight calling etc.
well in the good old days drive testing was done all the time to find problems with handovers. Then the Ericssons of this world sold auto-optimising code so you in theory didn't have to do that (although their code didn't work).
Add in the fact that 3G-2G handover mid call has never worked and you get the problem that you are driving along and you can see the next tower but the phone refuses to hand over and drops the call instead.
GSMR has some interesting bits in it. but given the limited user base could be properly optimised.
Add in the fact that 3G-2G handover mid call has never worked and you get the problem that you are driving along and you can see the next tower but the phone refuses to hand over and drops the call instead.
GSMR has some interesting bits in it. but given the limited user base could be properly optimised.
Some of the early phone systems like our iDEN network let the phone choose the best tower. That is the preferred method, but forces the phone to do a lot more calculations which eat battery life, so those days are gone.
Most phones still constantly look for other towers, or other frequency bands on the same tower. This is called the "neighbor cell list" and is used to request a handoff to another cell.
Assuming that the tower you see is actually one used by the network that you are using, the handover situation is still much more complicated than having a signal. Every phone has a priority assigned to it. Ordinary paying customers are somewhere in the middle of the pecking order. Calls that have been progress for considerable time move down the priority list. First responders, government officials, phone company employees and others have a higher priority that the paying public.
Let's say you are driving and your phone looks to handoff to another tower because its signal quality is dropping. It sends out a handover request to the network, which then looks for a suitable landing spot that has available radio channels for a device of your priority. If there are no nearby radio channels available the network may try to hand you off to a less than optimism spot, or refuse the handoff and hope you are still there when a suitable channel becomes available.
Lets just say that I had a phone named Dolly. She had three identical sheep just like her. Dialing my number would cause all 4 to ring and a 5 way conversation could happen. Note, I was a phone engineer and the network knew what we were doing. Cloning is virtually nonexistent today since all phones can be tracked by their internal GPS devices and the modern LTE networks are very complex and secure.
As of two years ago my friend was still drive testing for Verizon. Towere are being updated with new capabilities almost daily and they require continuous testing. Much is automated in the network since virtually everything that happens in the cell base station is logged, but little is known about what happens in the phone itself. Anytime a new phone comes out from the Apple's and the Samsung's of the world a special test version is made and it is drive tested in several major city's problem areas. Ditto with any new IOS or Android release. The phone carriers are VERY paranoid about bad software or hardware bringing their network down.
Yes, and no. CDMA and everything based on top of it including LTE uses technology derived from some WWII encryption scheme known as frequency spreading. A voice conversation that requires 12.5 KHz of frequency spectrum is moving around across 10 MHz constantly according to a predetermined spreading code (Walsh Code). It may also be turning off and on for short bits of time depending on how much data is being sent.
So technically every AT&T band 12 LTE transmission occurs on a 10 MHz wide channel centered at 739 MHz, each chunk of data is spread out over all or part of that 10 MHz chunk of spectrum such that collisions between two competing chunks of data are very rare and taken out by the error correction algorithms. The same is true with voice communications, which are just streams of data. LTE divides that 10 MHz chunk of spectrum up into little variable width pieces. It also divides a channel up into several slices of time. A combination of time and frequency is called a resource block. At any given moment two phones may be on the same RF channel, but rarely if ever occupy the same exact slice of frequency and time.
George: Can we accept you know way more than me about specifics of US only cell systems and I know a bit about rest of the world 
. We have some major misinterpretations going on that would take at least 3 beers to resolve and don't want to pull this thread even further off course. And most would not care about the difference between WDCMA and COFDM anyway.
. We have some major misinterpretations going on that would take at least 3 beers to resolve and don't want to pull this thread even further off course. And most would not care about the difference between WDCMA and COFDM anyway.
It has been pretty obvious what happened from the start; cutting costs and corners. The evidence is piling up. Maybe we shall see some prosecutions next month.
- Status
- Not open for further replies.