While we're discussing wireless in general

[koffiejunkie]

I looked at a couple of manufacturers websites, but I couln't find an answer.

If I have a, say 108Mbits/s wireless access point, how does the available bandwidth get devided amongst everyone on that acess point. Or rather, how many people can connect to one access point and how much of the available bandwidht do each get?

Thanks

--
koffiejunkie

 

[nonroker]

Depends on your AP, but it's generally first come first serve. IOW. One guy can suck your AP dry

--
256k ... BAH..more like 25.6k
FSCK YOU Sentech!! [!]

 

[bdt]

koffiejunkie- 's not quite *that* simplistic, there's actually quite a bit to it.
Recalling info from a vendor WiFi course I attended recently, the elements that affect BW allocation include (but are not limited to- I'm not sure how much I'm going to recall here!):
- the presence of just ONE -B- device in the cloud (presuming all others are not only -G- but "super-G" ready) will bring the entire net down to B speed- 11Mbps.
- similarly one user (presuming all -G- users now) at excessive range will pull down the 'net as the AP throttles back to keep everyone happy. But this may be beatable, depending on yr brand as with some brands you can define cutoff speeds.
- "Super G" mode (the mythical 108Mbps) works by doing the same trick as dual ISDN- it bonds 2 channels. More fun here- in sunny .za we get 11 (of the 14 available) channels and you need to have 'em SIX channels apart (3 either side of the specific channel nr you're using) to have 'em not interfere with each other which means you'll need to use yr own mix of 1-6-11. IOW, *not* having 2 free channels (with free space either side to use) means you can kiss Super-G goodbye.
- then the AP dirty li'l secret (or at least not widely spoken about my manufacturers/vendors!): seeing as an AP has to deal with a packet TWICE, ie [user1] <--> [AP] <--> [user2] your effective throughput drops to half the max speed you thought you were going to get. Two WiFi NIC's in ad hoc mode, ie [user1] <--> [user2] don't suffer that BUT their NIC's are always shouting at full power seeing as they don't have an AP to calm 'em down and generally manage the link.

I reckon that's a start- it may even answer your rather vaguely *g* phrased question.. back to you!

EDIT- from the sublime to the ridiculous dept: if you're looking for range, you could do like the guys who managed to get 88.67km with unamplified Orinoco Gold USB WiFi adaptors that they connected into 2.9m satellite dishes! See http://www.wifi-shootout.com/ for some blurb.

 

[donn]

The WiFi shootout server gave me some trouble, but here is another link:

http://www.netstumbler.com/2004/08/04/wi_fi_shootout_in_the_desert/

Do you think that if we use 9ft dishes with MyWireless that it will make a difference? Say from 64k to 65k ? []

<hr noshade size="1">
Donn Edwards <font size="2"><div align="right">MyWireless: Diva-style reliability, dial-up performance (or worse)</div id="right"></font id="size2">
<font size="1">“If our government ever goes bad, as sometimes happens in a democracy... As we extrapolate our [surveillance] technologies into the future, if the incumbency has that political advantage over their opposition, then if a bad government ever comes to power, <b>it may be the last government we ever elect</b>.” See http://privacy.4mg.com </font id="size1">

 

[martin]

lol donn

 

[guest2013-1]

i've been advised by miro to rather go for an 11mbit AP than the 54 (or 108) mbit ones because of the amplification or something or other being better and more stable over a distance?

Anyone wish to comment on this?

Hell, my gran on a scooter with a memory stick is faster than Sentech's MyWireless!

 

[koffiejunkie]

thanks bdt. I guess I didn't know enough to formulate the question properly - sad, I know...

--
koffiejunkie

 

[bdt]

noone- it comes down to properly defining your requirement. If you're doing only short range, G-speed is great (ok, not cheap- but great!) If you're looking at distance- that's a whole different animal and G-speed just gets lost in the distance anyway so why pay for speed you can't use?

koffiejunkie- there's NO shame in not knowing something.. you should see how much of a n00b I am around Linux! But I CAN learn- and you can rethink yr need and rephrase yr question to get (maybe only a step) closer to what it is you're *really* after..

 

[guest2013-1]

well, im at one of megawan's former high sites, so i want to cover as much as possible with decent speed and be able to cater for quite a bit of people

Hell, my gran on a scooter with a memory stick is faster than Sentech's MyWireless!

 

[dorris]

you want my advice, go for a 54MB/s, I'm no wi-fi guru, but sense tells me its better,
reasoning,
on 802.11, your connection speed degrades rapidly with each step away from the AP.
ie if at one connection step away from ap, 802.11b gives 5.5 MBps, 802.11g might give 25MBps

if you trying to setup a mesh AP, realistically NOBODY on your mesh will connect at optimal speed, so the entire performance, will be highly degraded, that same degredation on 802.11g, will still be usable!

I agree, don't bother with turbog's etc, they hardly work anyway, and when they do, the advantage, is parallel upstream/downstream, ie 54 Each way.

 

[gripen]

look, i have to disagree. my experience with 54Mbps is bad. I have a 54Mbps WLAN running in my flat ie. close range. Not once have I seen a transfer rate above 1.8MB/sec. Nobody can say for sure that they get above 2 or 3MB/sec anyways.

the 802.11g is 25Mbps vc 5.5Mbps argument doesnt hold up. 'G' requires more TX power for the higher speed. Thus, 5.5Mbps requires less power meaning more range. Effectively there will be a point where using b or g will make no difference. ie. lets say with super duper antennas.. at 5km you will get 1Mbps no matter what you do - b or g will both be slow. then again there are so many factors at play that it is impossible to tell. one thing seems certain.. 54Mbps is not as fast as you would think and 11Mbps is more stable (ito of "advertised" speed) over distance.

ALL 802.11 is half duplex meaning one station only can send OR receive at one point. so your speed suffers horribly when you have even 2 stations on the same WLAN.

 

[gripen]

lets put it this way, I have a centrino 54Mbps (where I made sure I got 54Mbps) and a 54Mbps PCI card access point (I know, this could and most probably is the problem) but I am now converted. I think Im going to get a Senao 802.11b card with a decent antenna. That way I could see more access points. Speed helps jack when you have no range and most run B anyways.

 

[bdt]

dorris, a coupla things why super-G isn't optimal for noone's need:

1st- (802=x) x.11b uses a reasonably simple encoding scheme called DSSS (direct sequence spread spectrum)- it's not hugely fancy and as a result the actual modulated signal is still usable at some kind of distance. x.11g uses a significantly more complex encoding scheme (OFDM-orthogonal frequency division multiplexing) that falls apart quicker, when compared with x.11b at the same non-trivial distance. In user terms: x.11b can be used further out than x.11g when talking about any kind of long (greater than inter-office) distances.

2nd- it's physics, specifically the inverse square law of electromagnetic radiation (see http://en.wikipedia.org/wiki/Inverse-square_law). This is why TV/radio transmission works- they plant a huge antenna nice and high and pump a FAT signal up it- right next to the antenna you could probably cook food in thin air from the sheer signal power- but as you get some real distance (of the order of km's) it's still possible to get signal. Of course, with WiFi the power levels are lower AND it's a bi-directional link so the distances are smaller. But we counter signal loss imposed by distance by amplifying usable signal in the direction we want to point with using antennas that focus (no antenna ever amplifies signal, AFAIK) the available max signal to get more performance/distance.

And it's this phenomenon that greedyflyza's talking about when he says he gets craptacular range with x.11g, although I suspect there's more going on there than just insufficient signal! But that's an ENTIRELY different topic and I don't want to diss the man's setup!

Getting back to noone's stated intent- (disclaimer: this is thumb suck off the top of my head after coming in from a movie!) in the short term at least, a high-ish gain omni antenna is probably going to cater to his needs, though at R1324 retail price from Miro a 15dBi omni is a stiff investment (not forgetting it still has to be plugged into an AP via cabling)! But he can plug that into only one AP running x.11b and be on his merry way. Down the road, in order to cater to greater user density, the spectre of secotised antennas looms along with all sorts of other fun complications- but that's a story for another day.

 

[gripen]

My setup may be crap BUT it cost me R250. While its not running G speed it is supposed to..

Some tips:
- NEVER forget about the expense of pigtails. You will almost always need some kind of pigtail and they are pricey
- go for max power above anything else. 20dBm 802.11b is way way better than 15dBm 802.11g including speed
- keep cables and connectors minimised. 1m of LMR195 cable drops 0.6dB and each connector drops about 1dB. I can go into the science of free space path loss but take it from me.. 1dB is still significant
- an xdBi antenna would typically improve SNR by x dB's ie by clearing the signal from the noise. There is NO amplification.
- keep in mind that every 3dB is a double of signal power. 15dBi is a hell of a lot more than 8dBi

Enough of the engineering stuff.

noone, mate, dont muck around. Get a Senao 11mbps Access point / bridge / repeater. Then at least a 13dBi omni if you want people to connect to you. Or rather get a 45 degree 13dBi horizontal patch antenna - directional is always better.

 

[guest2013-1]

Was thinking about a 120 degree one as I'm rather high up. I can see Sentech's brixton tower from my place and I'm in roodepoort (thereby hillfox)

I'm buying a new car, so I'll have to save up for this...

Hell, my gran on a scooter with a memory stick is faster than Sentech's MyWireless!

 

[bdt]

greedyflyza, you mention a 1dB insertion loss for connectors- that'd be horrific if it were so! The N-Type product page at Amphenol (http://www.amphenolrf.com/products/typen.asp) mentions a typical value of 0.15dB loss up to 10GHz. To clarify the 3dB statement, too right- it's a power doubling alright- or half, going the other way- which is why we jump through all sorts of hoops to absolutely minimise that penalty, noone.

I'm gonna tease the bull by calling into question the notion of "better"- it really *does* depend on the desired usage *g*. That said, noone seems to want to paint only the sector in front of his house which makes sense.. the hill he's on (top of?) would make for a rather annoying radio shadow so why pump signal into it?

 

[gripen]

I wasnt sure about the connectors. Subject to proper workmanship etc. Just pointing (sic) out that it was a factor.

 

[keithr]

Reading the posts in this topic lead me to do some research on the TD-CDMA / CDMMA 2000 / 1xRTT technology as used in the IPwireless technology by Sentech.

The question to answer is following the above thread on the degradition of signal in 802.11, is it not possibly why or part of the reason why the 256/512 packages are not performing?

Has anyone with those packages AND signals greater than say 40% done some testing in the past 24 hours?

I suspect that signal quality (not only strength) has something to do with the poor performance of the higher speed packages (assuming of course that Sentech are not practicing some dumb kind of throttling).

 

[gripen]

actually signal shouldnt be a problem in 80% of the cases. Sentech has a license so they can use as much power as they need whereas 802.11 is power limited.

it is most probably wrong to assume that its not Sentech who are not supplying sufficient bandwidth. Ask an 6 month user and they will tell you. It can perform and it did. Its that simple.

 

[guest2013-1]

keithr, I had great speeds up and till they fixed the 256k users on the 16th of August... well, it's not fixed, and 512k isn't much faster than 128k

Hence my theory on sentech keeping their majority users happy, and keeping the 256/512 also down to 128k...

this is getting ****ing ridiculous, last night I needed to update a couple of our servers, but couldn't even connect due to timeouts. I'm requesting credit up and till I feel that things are fixed... I already paid a month where I had no real speed...

they're definatly losing me as a client as soon as iburst comes around...

Hell, my gran on a scooter with a memory stick is faster than Sentech's MyWireless!