Phil,
> I agree, and I frequently make the same point. The signal-to-noise
> ratio on a CDMA channel is dominated by interference from other
> users. (It's actually a signal-to-interference ratio, but the
> interference looks like noise). So in CDMA, many/most packet losses
> are actually caused by congestion -- just as they are in a
> conventional
> shared point-to-point link!
>
That's a good point! But I guess that the relationship between
offered load and loss rate would be different from that in a
shared point-to-point link.
> Congestion still has to be handled by the transport protocol, but
> channel noise is best handled by FEC at the physical layer, perhaps in
> combination with ARQ. I really don't think channel noise should be a
> concern of TCP.
>
Except that it would be desirable to throttle back the TCP sender
(i.e., do end-to-end flow control) when little/nothing is getting
through to the receiver without having the sender timeout/retransmit, etc.
> >Of course, it would be desirable to have the TCP sender quickly
> >ramp up when the link quality improves. Schemes such as the one
> >proposed by Brown and Singh (CCR, Oct 1997) might do the trick.
>
> I proposed an "ICMP destination reachable" message many years ago. Is
> this like that?
>
I'd have to go back and look at the paper to be sure, but off the top
of my head I believe they played tricks with the receiver-advertised
window. During periods of disconnection, an agent at the base station
would send back an ack with advertised window set to zero. This
would activate the sender's persist timer but prevent retransmission
timeouts. Once the link quality improves, the agent would restore
the original window, causing the sender to start sending again
(of course, there'll be restart-after-idle like issues to contend with).
A pretty intricate hack...
> There are two kinds of frequency hopped systems, slow and fast. I
> think even the slow systems typically hop at least once per packet,
> which means retransmissions will occur after a hop.
>
> If you hop faster than a packet time, erasure-correcting FEC is a good
> way to deal with hops that land on narrowband interference.
>
You may well be right. All I know is that in the ISM band, you are
required to hop only every 400 ms or so. Depending on the link speed
and the frame size, there may be room for multiple frames before
a hop is needed. But I guess it is also true most systems hop far
more frequently than the etiquette rules require them to. And I am
not sure what non-ISM systems do.
-Venkat
Venkat Padmanabhan
Microsoft Research
padmanab@microsoft.com
http://www.research.microsoft.com/~padmanab
This archive was generated by hypermail 2b29 : Mon Jan 28 2002 - 09:12:19 EST