At 11:17 AM 10/7/97 MET, Francesco Potorti wrote:
>Is this reasonable? What is fiber quality? Are we implying that tcp
>over satellite will only work for practically error-free channels?
>What about the Ka band, where atmospheric attenuation due to rain is
>common and most certainly causes long periods of degraded channel
>quality?
Most of the Ka-Band systems will provide better than 10^-10 for better
than 99% of the time. All the systems claim something in the order
of 99.5 to 99.9% availability. So that other .5 to .9% of the time
you would see bit error rates down to 10-7 or so. The final .1 to .5%
of the time the link would be almost entirely gone (in at least one
direction).
>
At 08:45 AM 10/7/97 -0400, Mark Allman wrote:
>therefore take the drop as congestion and backoff. TCP will work in
>a satellite environment with a high bit-error rate, just not
>particularly well. But, that is the case for any channel with a
>high BER. So, I think it is quite reasonable to _recommend_ using
>FEC for good performance. To me, it is the same as recommending the
>use of TCP SACKs. You will get better performance with FEC than
>without. However, it is not necessary for TCP to work.
Even with FEC you will get periods of time with high BER. Especially
with the Ka-Band systems. The nature of the satellite links will however,
be that BER will change rather rapidly when it does fade or come out of
a fade. The bigger question is, how fast will the protocol take to
recover from a short period of BER=10-2. Another interesting question
for the Ka-Band systems is what happens when it is the gateway that
faded (return links) and 10,000 or so TCP sessions are active! Then
you will get both congestion and fading losses on the outbound path,
and fading losses on the return link paths. Even if your TCP session
is remote to remote, you could be affected by larger variations in
bandwidth allocations (congestion accessing the satellite).
The fact of FEC is that it is used to reduce the cost of the service
(smaller terminal, less transmit power, less power used on the satellite)
rather than increasing the BER performance. The problem is then that
when the link begins to fade it goes from nearly no errors to nearly no
good bits very quickly, then back to no errors very quickly.
Ian McEachern
Ian A. McEachern (Direct) +1-604-320-2654
Director, Network Design (Reception) +1-604-294-0040
IMT ComSys (Fax) +1-604-294-5506
5284 Still Creek Ave. (e-mail) [email protected]
Burnaby, BC V5C 4E4
This archive was generated by hypermail 2b29 : Mon Feb 14 2000 - 16:14:30 EST