With several collaborators I have been investigating the impact of using a larger initial congestion window in TCP for some time now. The proposal is now a standards track RFC. Some pointers to our work:
(Posted: Fri Oct 25, 2002)