Slashdot Mirror
WiFi Woes With .11g
Herby Werby writes "The Register has an article on the incompatibility between .11g and .11b across differing unnamed vendors due to premature roll-outs. The part which really hurts is the suggestion that if there's a .11b participant to your .11g network then either it gets ignored or the network reverts to .11b status. Anyone tried this yet with their new Powermacs?" As the article points out, this is most likely due to the fact that .11g hasn't really even been set as a *standard* yet, so incompatibility is to be expected. I just hope vendors get really good with flash updates.
Little "(2)" at the bottom of the page says:
(2) Based on IEEE 802.11g draft specification. Data rates greater than 11 Mbps require an AirPort Extreme Base Station, an AirPort Extreme Card, and an AirPort Extreme-ready computer. To achieve maximum speed of 54 Mbps, all users must use AirPort Extreme Cards. Actual speed will vary based on range, connection rate, and other factors
"When one or more 802.11b users connect, the wireless network begins to decrease its maximum data rate to accommodate them. When many 802.11b users are active on the wireless network, the overall network data rate begins to approximate 802.11b rates."
See the "Airport Extreme Technology Overview" at the bottom of this page.
I wasn't able to find any cards out on the market that'll work under Linux.. Most of them use the Broadcom chipset (or a newer rev of the prism chipset) that doesn't work with the existing prism drivers out there.. =/ I tried to toy around with the Linksys "54g" card.. but.. ended up giving up and brought it back.
*shrug* I figure I'll just wait the 4 months until 802.11g is out of draft and is actually standardized.. People are saying there will be flash updates for the cards.. but.. *shrug* I don't trust word of mouth too much..
Just when you make it idiotproof, some idiot builds a better idiot.
I bought a Buffalo 802.11g Air Station for my house that delievered a 4x speed increase over 802.11b. When My wife comes online with her Apple laptop using 802.11b the speed does drop back to 802.11b but since we are hardly ever on at the same time this isn't a problem. I still enjoy the faster speeds and Buffalo has a guarentee that they will upgrade or replace all hardware to meet the standards when they become ratified. Till then i'm enjoying a preview and lovin it...
I thought 802.11g was made a standard according to this /. post.
Boredom's not a burden anyone should bear.
My 12PB works great running against my Buffalo Tech WiFi base station and high-gain attenna. How many people really need the faster throughput? 11Mbs seemed fine before the PB arrived with the extreme card and it feels fine now. How many people are buying this stuff to run directly against their DSL or Cable. Let's see... 54Mb/s right up to the goofy router and then, wham, back to 1Mbs/128Kbs down/up.
The advantage of the new PowerBook like isn't really with the "g" - it's with the great attenna placement. My signal strength has never been better.
I have a Linksys 11g Access Point. There are about 7 11b users (mostly linksys pc cards and usb adapters). I use a powerbook (12inch) with the 11g card, and everything seems to play nice in that network. The powerbook always connects at 54, though the acctual throughput never gets close to that (I test with iperf). The 11b clients don't seem to affect the 11g client.
"You never know when some crazed rodent with cold feet might be running loose in your pants."
-Calvin
The IEEE Wireless Standards Zone overview is here.
Recent news from the IEEE re: 802.11 is available here.
I want to drag this out as long as possible. Bring me my protractor.
This article has so many errors in it I don't know where to start.
7 61 for the complete story. The only sensationalism here is for ignorant reporters being used by vendors who want to inject some controvery into a very technical process.
It says that because of incompatibilities between 11g implementations the IEEE was "forced" to decide between them. WRONG!!! The decision on a final IEEE 11g standard has had the SAME EXACT schedule for the last year per the association's roadmap. And, as the article DOES NOT SAY, the IEEE gave approval to a draft last week, RIGHT ON SCHEDULE. See http://www.80211-planet.com/news/article.php/1584
The Register really botched this story. Big time.
This client decided to put a PC on the network along with several networked printers. All networked printers worked fine with current Macs utilizing the "802.11b part" of the Airport Extreme (802.11b/802.11g hybrid)
The PC, with a Linksys 802.11g card didn't like the network - while it saw the network and Macs saw it, no connection could be made to the internet via AE BaseStation t1 internet.
I called Linksys, and they said, "At the moment, the Linksys 802.11g (which includes a new implementation of 802.11b) was only compatible with other Linksys equipment". Phone support managed to help me get the 802.11b working. I was transferred to a tech where discussed timeframes and support. I was told that Linksys is actually working with Apple to make a standard since more people initially will buy Airport Extreme. I was told to expect a flash updater for both units by the middle of March.
Yell & scream & rant & rave... it's no use... you need a shaaaave ~ Bugs Bunny
What is 802.11g? It is a mix of 802.11b and 802.11a.
802.11b, the commonly seen version, runs at 2.4GHz, just like Bluetooth. Why? Because 2.4GHz is the natural resonance frequency of water, which is the frequency of microwave ovens. So 2.4GHz was left open years ago, because nobody thought it would be any use (a 1KW noise source could completely swamp out the 1nW power of a tranmsitted frequency).
802.11b uses a digital modulation scheme called CCK, which is basically a fluffed up version of QPSK. 802.11a uses a more advanced modulation format called OFDM, but at 5.2GHz. OFDM is better able to operate in an environment with multiple reflections, but requires a much more complicated modulator/demodulator. But the complexity gets about 5x the data rate in the same bandwidth.
802.11g was a higher data rate version of 802.11b. Texas Instruments had proposed to use a data format called PBCC to get higher data rates than the CCK used in 802.11b. Intersil proposed to use the OFDM from 802.11a. A standards committee war started, and the end was TI lost. TI wanted PBCC because it was already working on a chipset that would support it, giving TI a great advantage. Of course, Intersil waas probably doing the same thing. When TI lost, it tried to do an end-run around the standard by releasing its chipset anyway. The Dlink plus series and USR 22MB/s Wireless components use the TI chipset.
No, I don't trust in god. He'll have to pay up front, like everybody else.
I can't only speak from my limited experience, but we bought the Linksys 802.11g router and were unable to connect to it from any of our 802.11b cards (we've since bought a 'b' access point to hold us over for now).
Check out 3com's solutions. I think that they are linux compatible across the board. Everyone else just says "Winblows XP" or Mac OS (9/X). Good luck.
Rumor has it that the incompatibility is due to Linsys' use of a Broadcom chipset. Apparently, they (Broadcom) are 'new' to the .11x market and their backwards compat to .11b is a problem.
Linksys being what they are will probably fix it with firmware then everyone will forget about it.
being ratified or anything like that. Regardless of the fact that 802.11g is in Final Draft status and will likely have very few, if any, changes... the issue is entirely whether or not 802.11b vendors completely implemented the 802.11b standard.
802.11b / 802.11g compatibility relies on implementation of RTS/CTS in the respective stacks. Many 802.11b vendors failed to implement this (for whatever reason).
Further, the idea that 802.11g access points revert to 802.11b when as little as one 802.11b client is present is a myth! What happens is that when an 802.11g access point is run in compatibility mode, it is forced to use RTS/CTS. The data rate is not slowed down. Rather, it experiences slightly greater overhead as RTS/CTS packets must be used.
Let us take the time to stop all this FUD now and educate ourselves.
In this article, EE Times looks at some of the same issues.
.11b and .11g basestations, Krewell said, clients will automatically default to the lower 11-Mbit/s bandwidth of .11b.
.11b chips for its Banias notebook platform, following that up with .11a/b combo chips within three months, and probably add support for .11g by the end of the year, Krewell said.
However, some of the same misinformation prevails:
And in an environment of mixed
Not true on two counts.
First, the only reason that a/b access points don't do this is because they're basically two different access points in one box! If a b/g access point had essentially two access points - one b, and one g - within itself, it wouldn't need to scale back either! Which brings me to...
Second, g clients don't scale back to 11 when b clients are present. They will get slower, but only because of the way the packets are interspersed. When 802.11b is present on an interface where g is present, everything, including b clients, will slow a little bit; by about a third. But g clients will not slow to 11.
Also, Apple's equipment has the ability to force b or g only, if needed in a particular installation.
Ultimately, one Apple design manager said, chip sets will support all three WLAN standards, eliminating any conflicts. Indeed, Intel intends to initially ship
Looks to my like it'll be a wash in the end, and I'd rather have g, albeit a draft g, right now (which, if there are any changes, will most certainly be updated to the final g via a firmware update). I can still connect to all b access points, and have increased speed when connected to my g access point (connected via 100mbit ethernet) today.
Note: this was posted wirelessly over draft 802.11g-Draft6.
Maybe Broadcom's Jeff Abramowitz won't tell which legacy cards have a problem with 11g access points, but slashdot readers should have no such constraints. I was using an Apple airport with an Orinoco PC24E-H-FC laptop card. A week ago I installed a Linksys WRT54G. The wireless part (with first firmware release) wouldn't talk to the Orinoco card. Installing the January firmware revision solved that problem. Also, the article suggests that using a large number of 11b clients might diminish the results for the 11g users. Setting the new 11g access point to a different channel (frequency) and running both at the same time, with 11g clients using the new channel, should solve that problem. (Assuming one can specify the channel on which the clients are to work - as in Linux).
You should read this link.
This article for a review of the Linksys G device:
Toms Hardware Review
Lot's of possible headaches listed.
I bought one anyway, since my SMC Barricade Router broke down the other day. I could have bought an A or a B, but since I try to hang onto my equipment as long as possible I decided to risk it by going with a G machine.
I don't have any wireless client machines yet, my house has plenty of cat 5 in it already, so I cannot attest to Tom's review.
BTW, I do not recommend SMC, their device was constantly overheating on me. It's just not acceptable to have to walk all the way to the other end of the house when I want to use the internet (i.e. to turn on/off the stupid Barricade router).
Caution: Contents under pressure
When my wife's powerbook arrived, we plugged it in and got it on with the extreme basestation. My older powerbook connected through it in mixed mode (or whatever that is). Her old Dell laptop couldn't connect to it. Later we were able to get the dell connected by switching to 802.11b-only mode, but that's lame.
:)
Anyhow, while it was in b/g mode, her laptop was able to connect and copy files (via AFP) much faster than my powerbook. The "server" was an iMac with 100mbit ethernet connected to the extreme basestation via a switch. I didn't do any throughput testing, but it was noticably faster.
As for achieving "maxiumum speed of 54Mbps, all users must use Airport Extreme Cards", i think it makes sense. Lets say you have 5 participants in the wireless network (4 clients and a base, or whatever... its not terribly important in this high level example). Lets say one of the clients has an 11b card and the rest have 11g cards. Then lets start transferring lots of data to and from each of them.
At equilibrium, I would expect that each client will be using about 1/5 of the available spectrum. So the 11b client can get about 1/5 of 11mbps, or 2.2mbps. The others could theoretically split up the remaining spectrum and achieve about 4/5 of 54mbps maximum, or about 41mbps. Even though the 11b client is only sending 2mbps, it takes up 10mbps-worth of spectrum to do it, so the network as a whole can only achieve about 43mbps, not 54.
Hopefully that makes sense...
blog
If the orinoco card doesn't find the network at all, make sure the advanced wireless settings of the router have "preamble" set to long.
If that fails, you'll most likely have to change the maximum data transfer rate to 11mbps, thereby hamstringing your 802.11g router.
Craenor
According to this Apple Knowledge Base article, the speed of the base station DOES NOT throttle when 11b users are connected.
It specifically says:
"Mixing clients on an AirPort Extreme network
When you mix 802.11b (AirPort) and 802.11g (AirPort Extreme) clients on an AirPort Extreme network, each type of client receives an appropriate data throughput rate. The 802.11g clients continue to receive data at a higher rate than 802.11b clients.'
The "little 2" is probably there so people don't think that when an 11b user is transfering files to an 11g user that the transfer will zoom along at 54mbs. In this scenerio, all user need to be extreme to get high speeds.