🚀 go-pugleaf

phr@telebit.com (Paul Rubin) writes:

>A co-worker of mine brought up an interesting idea.
>He thinks it might be very difficult to intercept data
>by tapping a phone line of someone using a modern high
>speed full duplex modem.  The reason is that the full
>available bandwidth of the phone line is being used
>in both directions; the modems separate out the signals
>with echo cancellation, which means that the modem
>takes the phone signal and subtracts what it itself
>is sending--the difference is what the other modem
>is sending.  But a wiretapper might have an awfully
>hard time of separating the two signals.  

Well, I can see how it'd be hard to tap at one point with
a high impedence feed and then separate the data.  But
I can see differential methods working.  Such as putting
a resistor inline and using the differential and comon mode
voltages to determine the amount of signal coming from
each direction.  I suspect that you could use transmission
lines to provide time delays and extract the signals
based on lightspeed delay.

I think it'd be more difficult though to compensate for
the echos in the system.  The V.32 or V.32bis modems on the
ends have their own training sequences to detect echos.
a tapping modem in the middle would have to be able to
compensate for the echos based on a handshake it had no
part in making.  Which I think would be a much harder
DSP problem than the normal manufactor has to cope with.