On Tue, Aug 05, 2003 at 01:26:06PM -0700, Patrick Leary wrote: > The article makes seems to make the case that this has something to do with > CSMA/CA. It does not. I has everything to do with having fallback rates (a > good thing in WLAN) and adaptive modulation. This issue can be experienced > with polling, scheduled MAC, etc. as well. It is why smart network builders > prefer tight cells with most client links to any given AU achieving the > highest order modulation. It is also how you can tell the novice who has not > yet learned this issue - he/she is the one bragging about having client on a > PMP network 10+ miles out. Such a person has not yet learned that such a > client costs them as much a 10x the resources of the AU as a client in the > 11Mbps cell of the same AU (assumes an 802.11b network).
Patrick, It does not have a little to do with CSMA? Say that 802.11 stations A, B, and C each produce frames with the same size distribution. A is associated at 1Mbps, B at 2Mbps, and C at 11Mbps. Each of A, B, and C has an equal chance of accessing the medium under 802.11's medium access rule. So A, who sends similar-sized packets as B and C, at a lower data rate than either, dominates the network "air time." (It is more complicated than that. The contention windows can shrink and grow. I have a hunch that fast stations' contention windows grow longer than slow stations' when the network utilization rises, exasperating the problem. Does anybody know?) Under another medium access rule, say TDMA, A, B, and C have equal air time. Say that they use the same data rates as above. It seems to me that station A cannot dominate the air time; actually, it will lose pretty badly in a TDMA system regardless of the utilization. Dave -- David Young OJC Technologies [EMAIL PROTECTED] Urbana, IL * (217) 278-3933 -- general wireless list, a bawug thing <http://www.bawug.org/> [un]subscribe: http://lists.bawug.org/mailman/listinfo/wireless
