Re: How big would an SSTO be?

On 2 Jul, 13:15, Alan Anderson <arand...@xxxxxxxxxxxxx> wrote:
Ian Parker <ianpark...@xxxxxxxxx> wrote:
1) There are many examples of secure networks where the data is
encrypted. There have been no examples of cases where encrypted data
has been cracked by hackers. The basic problem lies in the way data is
handled in the current Internet.

"The magic words are squeamish ossifrage." That example is from 1994.
See also "DVD Jon" and DeCSS.

2) The basic problem with hacking is that malware is allowed to get
onto your computer. In an ATC system all software used would be
present at take off. The operating system (on a ROM) would not allow
software to be downloaded except though special ports not accessable
by radio.

Malware is only one facet of hacking. The more fundamental issue is the
matter of authentication. The original computer hackers found ways to
get inside systems and manipulate them. That sort of thing would be of
primary concern with a computerized aircraft pilot. You don't have to
install any software on the flight system. You just have to convince it
you're a valid source of commands.

Authentication is quite easy. The RSA protocol allows you to have an
encrypted signiture. You need some knowledge on both computers that
they and no one else shares. These can be verified secretly by the RSA
protocol.
3) Details of tall buildings and high ground would be in the local
aircraft database. If it was asked to do something that was prohibited
by the database it would ignore ATC.

"3" has in fact already been implimented on a number of aircraft.
Attempts to crash them result in the removal of control from the
flight deck.

Attempts to *fly normally* have also been known to remove control from
the flight deck. One Airbus resisted pilot input to the point where it
caused a crash when it was accidentally configured for an aborted
landing.

This is quite an interesting point. What in fact caused the crash was
an interaction between the autopilot and the pilot. If the
configuration had been done from the ground this would never have
happened! When you are getting verbal instructions from ATC and
feeding these into the autopilot there is always the opportunity for
finger trouble.

There's a joke among pilots of such aircraft. The typical last words on
the cockpit voice recorder before a crash used to be "Oh, sh*t" or
something similar. Now they're more like "What the h*ll is the
autopilot doing now?"

It is interesting to note that the question now is less whether a
computer could cope with an emergency situation. In fact a Neural
Network can successfully fly aircraft that could not be flown by a
human. The question is of the reliability of the technology and
hacking.

Yes, that's a valid question. The best answer at present is that
removing people from the loop is rarely a good idea. Note that the
definition of "emergency situation" has to be made in advance in order
for a computer to deal with it. The problem comes when unanticipated
situations arise. Reliability isn't the issue in such cases.

All emergencies have to be preprogrammed. This is true. We look at the
chances of malfunction of aircraft systems, we have a drill for the
failure of any component. We look at adverse weather conditions etc.
Now if the system is really being put in a situation where it cannot
cope, it can always request remote piloting. In practice this request
would be extremely rare. There are of course irrecoverable situations
where nothing wil save you.

- Ian Parker

.

Quantcast