Re: Satellite Cellphones

[die@pig.jjm.com (Dave Emery)]

> 
> 
> Whats going to happen when direct satellite phone service becomes a
> reality ( as in the joint Loral/Qualcomm effort ) ?
> 
> When one notes that it is difficult to impossible to prevent service
> stealing in the cellular environment, how will it be prevented in
> direct broadcast?

	My understanding of both IRIDIUM and the Loral effort are that
the satellite will be used to directly connect a remote user to a ground
based MTSO (switch) which will actually route the call out over land
lines or another satellite link.  This implies that the satellite is not
being used as a classical bent-pipe repeater linking the transmitter of
one satellite phone to the receiver of another, but rather as a space
born cell-site linked to the MTSO via a separate radio system on a
completely different frequency band not unlike the terrestrial microwave
links that link most cell sites to the current MTSOs. 

	This means that there is no way for a mobile satellite user to
bypass the switch and use the satellite directly to relay his
communications to another satellite phone, just as there is no way in
the current analog AMPS/NAMPS cellphone system for a user on one
cellphone to talk directly to another cellphone without going through a
cell site relay and the MTSO switch.  Thus the switch can always serve
as a gateway authenticating users, and providing billing and access
control services. 

	Even in the more advanced concepts in which the satellite actually
does some measure of on-board switching between the mobile terminal
uplinks and downlinks, this switching is almost certainly going to be
controlled and managed from the ground even if the actual traffic path
goes only through one or more satellites.   And hopefully, someone has
already thought of the problem of a user of a satellite switched link
having his direct satellite link taken over by a higher powered bad guy
perhaps only for a few seconds to transmit a burst of data to
a confederate monitoring the downlink.

	Thus the problem of satellite phone access security is not very
different from the problem of current terrestrial cell phone access
security - namely that a remote user coming from somewhere out there
and preporting to be a legitimate subscriber requests a connection
over a very unsecure link subject to both easy monitoring and various
kinds of spoofing.   Obviously this is a very natural place to apply
cryptographic authentication technology such as zero knowlage proofs.

	As most of you are no doubt aware, the problems with fraud in
the current AMPS cellular system in the US are due to a very bad design
decision 15-20 years ago to not use some form of strong encryption of the
authentication exchanges between the mobile phone and the switch. 
Instead all of these take place completely in the clear, including
transmission of the ESN (electronic serial number) which is the only
trully unique ID a cellphone contains.  Thus anyone with trivially
simple equipment (basically a scanner and a PC and a very simple
interface card) can determine the ESN and MIN (telephone number) of
someone elses cellphone by listening in to its tranmissions and use
these to make fraudulant calls charged to that number. 

	Had the designers of the system simply cbc DES encrypted these
messages under a fixed cell or system wide key, most of this fraud would
have been too difficult for all but those with much better things to do
with their time and resources to attempt (assuming some elementary 
common sense in the design of the message formats to thwart simple
replay attacks).

	One suspects that these kinds of fraud in a cellphone system
designed using the concepts well known now for proper crytographic
authentication and resistance to spoofing and replay attacks would be so
rare as to be unimportant compared to other losses.  In such a system
security of the authentication information would no doubt be the major
risk, as the designers of the European GSM system seem to indicate with
their choice of a removable smartcard security module (token) that can be
carried on one's person.

> It might not be possible to detect unauthorized encryption schemes
> without direct monitoring. Do the Feds think this is can be won?
> 

	I don't imagine it is ever possible to detect unauthorized 
encryption without direct monitoring.   And for data transmission as
opposed to voice, this is a very hard problem since some considerable
fraction of data transmitted is such things as compressed binaries 
and images and things in unusual formats which don't lend themselves to easy
automatic recognition at low cost.

							Dave Emery