I think you guys know most of this already, but here is my take FWIW. I'm not a WISP, but I spent 5 years leading the design and development of an 802.11[agb] security system. We did our own polling solution based on 802.11e HCCA to solve the RTS/hidden node problem.
All things being equal (which they often aren't) 802.11b will give you a higher S/N and C/I than 802.11g, because in almost all cases and especially at higher speeds. 802.11g has to lower the PA power because of the PAPR of OFDM and meeting the 802.11g EVM spec. It is true that 2.4 GHz can be very polluted. We found the noise floor to be really awful. You would be surprised by the number of "entities" that know they are way over the FCC max power in 2.4 GHz, but I digress. We once measured over 300 PHY errors a second on an "unused" 2.4 GHz channel. The number went down to 150 PHY errors a second inside an FCC chamber, if you can believe that. Having said all that we didn't use 802.11b at all because it's data rates are too low for video. Also while we supported 2.4 GHz, we mostly deployed at 5.8 GHz ISM because of the increased power available there and the pollution was much less, but that maybe different now. For 802.11[ag] mutlipoint, the sweet spot speed wise is 18-36 Mbps. It's very hard to keep a multipoint system at 48 or 54 Mbps because you need a great deal of link margin and with all cards you loose power as the speed increases to maintain PAPR/EVM. For point to point with direction antenna relief you can often maintain 48 or 54. Antennae make a big difference, as others have noted horizontal polarization is usually best and make the beam as narrow as you can afford because it raises the effective gain. However, if you are in an area where everyone else is horizontal it can make sense to try vertical. With some of the antennae we used that was as simple as rotating the antenna 90 deg at both ends. Watch out for crappy antennae, cheap cable, bad connectors, and so on. That can often cost you a few dB. In the product I designed I spent more time then I care to admit trying to make a very tough loss budget that I set out as a goal. There is no substitute for link margin, you can never really have enough. I can confirm that our sweeps with a spectrum analyzer show lots of opportunity to use 5 and 10 MHz channels, as others have also noted. For WISPs it would be "nice" if chip vendors designed the radios so that you could set the channel bandwidth from 5-40 MHz in 1 MHz increments. It can be done but probably won't be, although maybe the Microsoft WhiteFI stuff force the chip vendors to do it. In WiMax and LTE they are already doing some things close to this. Still 5, 10, and 20 isn't bad and probably hits the sweet spot or 80/20 rule. One of the down sides of fitting a 5 or 10 MHz channel in a sweet spot is that it can change at any time. Best, leb At 9:58 AM -0500 10/1/09, Jason Hensley wrote: >In 2.4 land, if you have a lot of noise, which protocol is better - B or G? >Is it better to run an AP as locked into one mode or is it OK to do a mix? > >Max I want off of 2.4 customers is 3meg so not that worried about the extra >speed that G will provide, but, I would like to know which is more stable? >I've always thought that B was more stable overall but just provided less >bandwidth. I've gotten some info that may counter that. What's the >real-world experience with folks in a high-noise environment, combined with >a higher useage AP? > >I've got an AP that we've run in B mode only for a while. We've started >having problems with it - speeds go from 3meg at the customer to 200k and >fluctuate constantly. We've worked with RTS, ACK timeouts, etc etc and >nothing seems to have improved the stability. For testing purposes we put >up another AP right next to the one we're having trouble with. Switched two >of our gaming clients to that one (setup as G mode only) and they seem to be >doing better, but not quite as good as we feel they could be. This is on >Deliberant AP's (Duos). The backhaul part of it is not the issue - we can >pull close to 15meg back to our office when cabled into the AP. We have >other Deliberant APs that are running MANY more clients than this one so we >know it's not limitations of the equipment. AP is on top of a water tower. >Have taken all clients off and brought them back on one by one and it did >not reveal anything significant. With just one customer on the AP started >acting up again. Swapped radios in the AP thinking we could have one going >bad and still no luck. > >2.4 antennas are H-pol. We have a ton of noise in the area, but we've been >through basically every channel and it did not help either. Other AP's in >the vicinity are performing fine. Thought of the multipath issue so we >raised our test AP up a little higher than the other one. As I said, the >test AP seems to be better, but next to it on top of the tower we can get >around 8 or 9 meg down (locked into G mode), but at the CPE's we're still >barely getting 2.5-2.8meg. > >Any thoughts? We changed everything we can. The new "test" AP has a 9db >antenna compared to the 13db on the "production" AP. Other than that, they >are identical as far as equipment goes. > >So, back to the subject question though, what's real-world experience with >G-only mode in the field? > > > > >-------------------------------------------------------------------------------- >WISPA Wants You! Join today! >http://signup.wispa.org/ >-------------------------------------------------------------------------------- > >WISPA Wireless List: wireless@wispa.org > >Subscribe/Unsubscribe: >http://lists.wispa.org/mailman/listinfo/wireless > >Archives: http://lists.wispa.org/pipermail/wireless/ -- l...@iridescent.org -------------------------------------------------------------------------------- WISPA Wants You! Join today! http://signup.wispa.org/ -------------------------------------------------------------------------------- WISPA Wireless List: wireless@wispa.org Subscribe/Unsubscribe: http://lists.wispa.org/mailman/listinfo/wireless Archives: http://lists.wispa.org/pipermail/wireless/
