On Monday 27 April 2009, Mike Kershaw wrote: > > > Construction can definitely influence wifi. 2.4ghz doesn't > > > care much for any walls, definitely not metal, wire mesh, etc. > > > if the pumps, etc are anywhere in between you and your AP it > > > won't go terribly well. > > > > Right -- the problem with reflective objects is the problem of > > "multipath" -- the reflections of transmitted signals add > > voltage-wise and small time delays from the increased path length > > of the reflections means that they add in and out of phase. Two > > equally- > > Wifi is relatively resilient to multipath, and in the case of > 802.11n, *requires* multipath to get the full data rates. (They do > crazy ninja signal reassembly using multipath).
I don't know about hardware internals of 802.11n, but I understand HD-tv deals with multipath by using several variable delay lines in order to sum up multipath signals; I imagine 802.11n hardware could use a similar trick. HD-tv doesn't have much option but to do it, because otherwise the video image gets turned to mush, and the same is probably true of high data rates of 802.11n. i.e. I think it's one of those times that "necessity is the mother of invention". > > These antennas have an omnidirectional or "doughnut" propagation > > pattern to them that is perpendicular to the "stick" of the > > antenna. Signal level drops off as the angle deviates from > > perpendicular, and no signal propagates from the direction of the > > ends; you want to aim these perpendiculary to where you want to > > receive signal. For instance, if a computer that needs to use > > WiFi were in a direct horizontal path away from the router, then > > you want the "stick" antennas aimed straight up. If you want to > > receive WiFi downstairs, then you probably want to aim these at > > an angle. > > Generally speaking the antennas that come on an AP are about 2dB > gain making them rather spherical radiation patterns; there aren't > too many dead areas at that gain. You're probably right, but 3 dB means double the gain of 0 dB, so 2 dB isn't nothing. The standard spherical radiation pattern of 0dB gain is called "isotropic"; 2 dB = 2dBi, both mean "2dB over isotropic". I would have preferred if they had named it "spherical" because that gets the idea across better, but unfortunately most of the antenna mathematics were developed by physicists rather than engineers, which explains why the terminology and methodology is a bit backwards in some respects. > > I haven't needed to try this myself, but it looks like these > > antennas can also be modified cheaply: > > Your 11b/g/a APs only use 1 antenna - the second, if it even has > one, is in passive diversity mode (to combat multipath fade). You > can kind of mess around with them and it won't care overly, though > drastically confusing the diversity antenna isn't going to win you > any great benefit, if you're going to hack a homebrew antenna onto > it or start taping foil to it, disable the diversity antenna and > only use the one. > > In contast, 802.11n uses all 3 antennas actively, and cares quite a > bit more about them. Slapping reflectors and weirdo antennas on it > is almost certainly something you don't want to be doing; This is > exacerbated when you realize many 11n implementations are actually > dual-band 2.4 nd 5ghz, meaning any antenna mods you make have to be > clean on both frequencies. And if your AP is NOT dual band, you > probably won't see full 11n speeds, many vendors, including intel, > simply cannot do full 11n speeds on 2.4ghz and probably never will. > > So, simply put, don't go hacking up the antennas on your 11n AP. > You probably won't appreciate the results. That's probably good advice. At the same time, though, it is possible to make dual-band antennas, at least in general. You can usually tell whether a monopole is a dual-band monopole if there's a loading coil at about the midway point of the antenna whip. I know they sell aftermarket replacement 802.11 antennas with higher gain, but I haven't looked to see if they sell aftermarket ones for 802.11n. -- Chris -- Chris Knadle [email protected] _______________________________________________ Mid-Hudson Valley Linux Users Group http://mhvlug.org http://mhvlug.org/cgi-bin/mailman/listinfo/mhvlug Upcoming Meetings (6pm - 8pm) MHVLS Auditorium Mar 7 - Web Hack-a-thon - SUNY Newpaltz Apr 1 - EC2 and Cloud Computer May 6 - TBD
