Thread: [Madwifi-devel] BSD branch and Ad-Hoc
Status: Beta
Brought to you by:
otaku
From: John C. <jc...@me...> - 2005-04-22 18:21:09
|
I cvs'd the as-of-20050421 BSD branch of madwifi. The resultant driver seems to do merging and seems to run for a 'long period of time'. I have an experiemental setup consisting of 4 nodes, 3 of which have two 802.11g cards, and one has only one card. Each two card node is using channel 2 and channel 10 respectively. I've looked at a spectrum analyzer and it seems this has 'pretty' good channel separation. In a run over night with two 'netperf' tests in operation, on each of the channels, I had some rather different results. On one pair, the plot of the Mb/s seems to have a fast kickoff value, say 20 Mb/s, then shortly thereafter drops to about 10 Mbs. During the run there were somewhat 'erratic' points dropping to 'very low', in sort of a random patter. However, occasionally I'd see a drop to about 5 Mb/s, but then after some period of time, the bit rate would return to about 10 Mb/s. In the other pair, the kickoff bit rate was about 15 Mb/s, then dropped to 10 Mb/s and finally for a very long period until the end of the test, hovered around 5Mb/s. There were also points were the bit rate shot up to 10 Mb/s but these were random and infrequent. From the spectrum analyzer, it would appear that the area I'm in is pretty clean relative to other 2.4 Ghz signals, so unless there is some radar sweep in the middle of the night, the burglers used the Microwave, or similar, it is not clear to me why there's this variation in the bit rate. The iwconfig stats, don't seem to indicate horrendous errors either, on the pair which had a consistenly low bit rate over most of the test period, approx. 5 Mb/s, there have been only 6 'excessive retries', and 6 'invalid misc.' errors reported. This would definately not explain the low bit rate. The question to the readership, is 1) is anyone extensively testing Ad-Hoc, 2) have similar behavior been observed. Thanks John Clark |