On Thu, 24 Feb 2000, Alioune THIAM wrote: > Did anybody know why FullTcp don't work correctly with the Satellites > network ? I am trying to build a simulation like that provided in > sat-teledesic.tcl by changing the the udp agent with a FullTcp but there > always an error like : FullTcp::recv(_0345) Bad ack(21) to my Syn(1). > FullTcp::recv(_0346) god > packet lacking ack(21). > Thanks for your help. > Alioune (Tom, not sure you're still(?) on ns-users.) Alioune's right, and I'm kicking myself for not trying this earlier. I've only been using FullTcp with my static topologies, after playing with one-way tcp briefly over your constellations and deciding that my results were very uninteresting and I could get equally dull results faster by assuming the constellation didn't move for the most part, since looking at any transients raises an awful lot of uncontrolled implementation details-type questions. Attached is a tiny tarball duplicating what Alioune reports, run on a snaphot of this week; I've modified sat-iridium.tcl to create sat-iridium-fulltcp.tcl, and there's an errorlog and the minimal tracefile showing the sender exponentially backing off attempts since it doesn't appear to like what it's getting back. I'm not seeing the 'god packet' errors Alione reports though. If Alioune and I are going wrong in exactly the same way, I'd like to know what it is... any ideas? thanks, L. <[email protected]>PGP<http://www.ee.surrey.ac.uk/Personal/L.Wood/>
tar.gz tarball of fulltcp iridium code