Hi,
I have been conducting some analysis/ns simulations with TCP over typical
satellite links (LEO and GEO). I have come to the same conclusions as
Anurag's. However, my results show that the throughput is worse when the
average fade duration is close to the packet transmission time (which is a
function of the packet size and the bit rate) rather than the RTT (and
hence the IETF recommendation regarding using large segment size is
still applicable for this problem).
Regards,
--Hussein.
>
>
> ----- Begin Included Message -----
>
> >From [email protected] Mon Apr 5 18:36:22 1999
> X-Authentication-Warning: raptor.CS.Berkeley.EDU: tomh owned process doing -bs
> Date: Mon, 5 Apr 1999 18:39:53 -0700 (PDT)
> From: Tom Henderson <[email protected]>
> Reply-To: [email protected]
> To: [email protected]
> Subject: Re: BER and TCP/IP performance (fwd)
> Organization: UC Berkeley Computer Science
> X-Url: http://www.cs.berkeley.edu/~tomh
> MIME-Version: 1.0
> Content-Type: TEXT/PLAIN; charset=US-ASCII
> Content-Length: 2508
>
> Murat,
> I came across this message on the tcpsat mailing list today and thought
> you might be interested in these references concerning TCP modelling for
> non-IID loss channels.
>
> Tom
>
> ---------- Forwarded message ----------
> Date: Thu, 1 Apr 1999 10:23:03 +0530 (IST)
> From: Anurag Kumar <[email protected]>
> To: [email protected]
> Cc: [email protected], [email protected], [email protected],
> [email protected]
> Subject: Re: BER and TCP/IP performance
>
> With reference to Aaron Falk's mail excerpted below:
>
> Recently several analyses have been published
> that account for correlated losses (for example, as would occur
> in wireless fading channels). The results are quite different from
> those obtained for independent and identically distributed losses.
> As would be expected, in addition to the *average* bit error rate,
> the lengths of the good and bad periods of the channel become important.
> It turns out that rapid fading and very slow fading are both better
> than fade rates comparable to the RTT. If the link layer is always
> able to recover from losses, then essentially it is as if TCP sees
> a bottleneck link with a time varying bit rate.
>
> My papers on this topic an be found at http://ece.iisc.ernet.in/~anurag
>
> There are other papers by Zorzi & Rao (CISS 97 or 98 ?), Anjum and Tassiulas
> (Sigmetrics 99). And I am sure there are other papers too.
>
> -- Anurag
>
> On Wed, 31 Mar 1999 [email protected] wrote:
>
> > Jamshid-
> >
> > Thank you for raising an important point. Not all bit errors are equivalent.
> > Most (all?) satellite links utilize FEC. Block coding results in lost blocks
> > when uncorrectable errors occur. The packet loss rate due to corrupted blocks is
> > much different that that due to a uniform distribution of bit errors. So, the
> > translation of channel BER (after FEC) to packet loss rate (which is necessary
> > to interpret the results of the reference below will depend heavily on the
> > adaptation of IP to link layer.
> >
>
>
> _______________________________________________________________________________
> Anurag Kumar |e-mail: [email protected]
> Professor |Phone: (+91)-80-334 0855 or
> and Coordinator ERNET Project | (+91)-80-309 2387
> Dept. of Electrical Communication Engg. (ECE)|Fax: (+91)-80-334 7991 or
> Indian Institute of Science (IISc) | (+91)-80-334 1683
> Bangalore, 560 012, INDIA |http://ece.iisc.ernet.in/~anurag
> -------------------------------------------------------------------------------
>
>
>
>
>
>
>
> ----- End Included Message -----
>
>
This archive was generated by hypermail 2b29 : Mon Feb 14 2000 - 16:14:54 EST