[net.ham-radio] more thoughts on secure packet radio networks

ad7i@hou2h.UUCP (Paul Newland) (11/16/84)

i too have been thinking about how to verify that a user of
a packet network is who he or she claims to be.  i see from
laurens comments that our ideas are very similiar.  mine
are still somewhat half-baked but i will present them here
so that others can recommend improvements or alternatives.
 
what i am considering is the situation that a user would
like to change some parameters of a remote station (digipeater,
host, weather stn, nuclear reactor, etc.) but doesn't want the bad
guys to screw it up.  the though would be that when i was connected
to, say, my remote home station and wanted to turn on the the coffee pot,
the home station would respond with a command prompt that included
a random number (64 bits printed as a 16 digit hex value).  I would
encrypt that random number using DEA (that's Data Encryption Algorithm,
the software approach to DES) using the key-of-the-day and return
it with my command to turn on the coffee pot.  my home station would
also encrypt the random number it generated using the same key and
compare the result with what i sent to it.  if they are the same 
then it is very likely i am who i say i am (within the 'bad-guys'
fudge factor) and the home station will switch on the coffee pot.
if they don't match, the home station will give me a new prompt
and a new random number to encrypt and send back with my command.
note that the command and response are not encrypted, only the
verification data.

because every prompt from the home station would include a NEW
random number and because the keys are changed daily, it would
be very difficult for a 'bad-guy' to masquerade as a 'good-guy'.
Simply repeating the command and the encrypted random number
wouldn't solve the problem for the 'bad-guy'.  more sophistocated
techniques would be needed (and i am sure the 'bad-guys' would
try them -- some might succeed).

If you don't like DES/DEA then choose another encryption scheme.
At the least, DES is a known algorithm that offers good security.

what do others think about this???
cu, paul, ad7i

SK