Dear Vlad and Eddie,
The problem is that voice packets are much smaller than the typical maximal
segment size used by TCP (either determined by Path MTU or set to some
arbitrary value by the vendor). A TCP implementation that disables Nagle's
algorithm can therefore send packets totaling 4380 bytes of payload per RTT.
The actual bandwidth consumed is even higher, due to accumulated headers.
Precisely. I brought this issue up with Gorry 2 days ago.
I had reservations on using MSS to calculate the initial rate for TFRC
for the reasons you have mentioned in your email.
W_init = min(4*MSS, max(2*MSS, 4380)),
I "felt" it was better to use the packet size "s" instead of MSS to
calculate the initial rate. The problem of using "s" (I presume) is
on how you determine "s"? But I guess we "could" use the size of the
"first" packet that is being used to calculate X of 1 packet per
second. Correct me if I am wrong.
Regds
Arjuna
--
Dr.Arjuna Sathiaseelan
Electronics Research Group
University of Aberdeen
Aberdeen AB24 3UE
Web: www.erg.abdn.ac.uk/users/arjuna
Phone : +44-1224-272780
Fax : +44-1224-272497
