802.11 stuff

Sure. I have 3 customers doing exactly that. 2 businesses and a coffee shop. The trick is to use different channels (1, 6, and 11) so that they don't interfere with each other. If you try to handle both

802.11b and 802.11g on a single access point, the 802.11b users will slow everyone else down. However, with 2 access points, with one set to 802.11g speeds only, there's no slow down. There's less of a need for this today, where just about everything that enters the business of coffee shop can do 802.11g. However, that was not the case 3 years ago, when 802.11b devices were still appearing.

Well, do you have any 802.11b only client computahs or devices? If not, then just turn off 802.11b and set it for 802.11g only. If yes, setup a 2nd 802.11b device as an access point, and turn it on only when necessary.

Incidentally, if you plan to shovel 16Mbits/sec over wireless, methinks you're going to have to do some careful benchmarking and shopping. Many routers will not pass 16Mbits/sec. See chart at:

These are for wired, not wireless connections. Wireless only makes it worse. With 802.11g only, running at the maximum of 54Mbits/sec data rate, the best you can do is about 25Mbits/sec thruput. That's under ideal conditions, which only exist in the advertisers imagination. Add a little interference, a few reflections, some multipath, a dash of frequency selective fading, and your theoretical thruput become much worse. If you're going to do benchmarking, see iPerf and Jperf.

For a major speed increase, think about trying MIMO. Here's where the specs really get ugly (and everyone lies). In theory, you can do up to about 400Mbits/sec throughput, but nobody has been able to demonstrate that. Here's 300Mbits/sec:

I've seen 225Mbits/sec, but only in a rather unrealistic laboratory arrangement. That was at a range of about 8 meters. If you try going through a wall or two, your speed will drop back to 802.11g speeds and below. More realistically, you might get about 50Mbits/sec thruput, which should be sufficient for your 16Mbits/sec connection.

Incidentally, the 40MHz bandwidth flavor of MIMO (802.11n) disables

802.11b legacy compatibility, so you won't have to worry about that slow down. You can also disable 802.11g compatibility, but that results in some rather bizarre range versus speed effects.

Also, DOCSIS 3.0 is probably coming. Comcast and others are claiming speeds up to 50Mbits/sec sec. You'll need a new cable modem (Motorola SB6120 and SBV6220), and some care in selecting your router for thruput performance.

Reply to
Jeff Liebermann
Loading thread data ...

There's no requirement that you have all that initially. You can start out any way you want with minimum hardware and support software. However, you'll find that it's more and more difficult to add this stuff later, especially if you pass the costs on to the customers. For example, if you don't use bandwidth management, and a few customers get used to hogging the entire 16Mbit/sec all night, you're going to have problems with them if you suddenly decide to cut their available bandwidth. I've found it best to start off with tight security, draconian restrictions, aggressive abuse enforcements, and intentionally limited bandwidth, as it's much easier to remove restrictions, than to add them later.

Full disclosure: I'm no longer involved in the WISP thing. The problem was that very cheap DSL arrived and we couldn't compete with wireless.

Yep, I remember (but not the specifics). Bandwidth management simply means assigning a bandwidth limit to individual customers depending on how much they pay per month. Plenty of ways to do it in the router. Please note that I said "bandwidth management" and "abuse detection". I did not say "bandwidth throttling".

Abuse detection is really quite simple. Sniff for viruses, worms, spambots, and other malware. If detected, block the account until the user calls support (that's you) to ask what happened. Also leave a phone message or email. You're doing them a favor. Other forms of abuse are a bit more difficult to deal with. If you don't have the terms-of-service established defining what constitutes abuse, you should not other enforcing some random unstated policy. Set the limits, monitor the traffic volume, and enforced it uniformly. If they want more traffic per month, let them pay more. With simple traffic limits, you should have no problems with monitoring and enforcement.

It probably is too much for you to do all this. Just the log file reading alone is a major time burn. In effect, with wireless, you're doing everything that a land line ISP needs to do, with the added bonus of a marginally reliable delivery mechanism (which can be trashed by one leaky microwave oven). If you can't get cheap help (i.e. students), outsource the function to a network management company. Note that all the local WISP's have most of what I've mentioned.

By the way, congratulations for having remained in the WISP business for so long.

Reply to
Jeff Liebermann

Read it again. Take note of:

: One of the major differences between 802.11b and 802.11g is : protection, which is reguired... : : [...] : : Protection is activated whenever there is a need to ensure 802.11g : stations do not interfere with 802.11b stations. : : [...] : : Protection is activated when a non-802.11g station associates with the : network, as well as when non-802.11g-capable stations are transmitting : in the area. : : [...] : : By shunning stations that are not capable of 802.11g speeds, there is : an increased probability that protection can remain disabled. Of : course, a station that is shunned may very well find a different : network that can accomodate it on the same channel, which would still : trigger the activation of protection.

When you put the AP in 802.11g mode and turn off Protection, it will run in 802.11g mode without protection until the first 802.11b station in the area starts transmitting. Even when this 802.11b station ends up associating some another AP on a different channel, the station will have scanned all available channels and triggered the activation of protection on the 802.11g-only AP, halving throughput.

And when the AP has a "manuel switch" for activating protection, protection will stay on until the AP is rebooted (or the Protection "switch" in the AP setup is "toggled"). Otherwise, there would be no reason for having the Protection switch in the first place.

You speculate. How many 802.11b/802.11g APs have you actually compared?

Reply to
Axel Hammerschmidt

Quite a few--it's part of my work. How many have you tested?

Steve

Reply to
Steve Fenwick

Hmm! An 802.11 professional, unfamiliar with 802.11g 802.11b protection.

You've published your comparisons where?

Reply to
Axel Hammerschmidt

I'm not allowed--I work for a company that produces products (I do the HW, not the SW), so the results are proprietary.

Steve

Reply to
Steve Fenwick

Nope. Please read:

before blundering onward. Note that there are at least 3 ways to handle 802.11b compatibility when running 802.11g, none of which resemble your terse description. The original tests were performed in

2003, which suggests that there may have been additional algorithms or optimizations added in the last 6 years. The most common one currently is to turn off 802.11b compatibility completely when running various faster then "g" modes, such as Super-G, Turbo-G, and MIMO.

Also, you might want to look at the table of speeds in the FAQ:

or if you're a non-believer, the original Atheros article at:

Max TCP thruput with 802.11g 54Mbit/sec association is 24.4Mbits/sec. Max TCP thruput with 802.11g 54Mbit/sec association and 802.11b traffic present is 14.4Mbits/sec.

The problem is that 802.11b broadcasts beacons and preambles at

1Mbit/sec. 802.11g does it at 6Mbits/sec. When compatibility mode is enabled, the AP has to do it at BOTH 1Mbits/sec and 6Mbit/sec according to one algorithm. Others are a bit smarter and only broadcast at 802.11b rates after it hears a valid 802.11b packet.

Nope. I've done benchmarks using iPerf and jPerf showing that some chipsets are much better. Early 802.11g chips were disgusting. You didn't even need to associate with the AP to get a 50% slowdown. The AP would just sit there half the time listening for possible 802.11b traffic. Needless to say, benchmarks in 802.11g modes were slothish. You can see some of those on SmallNetBuilder. I'm too lazy to dig them out right now.

The last time I ran a test on my Buffalo or Linksys Broadcom chipset router running DD-WRT, there was almost no detectable slowdown when I associated with a WAP11 wireless bridge, but generated no traffic. Benchmarks with and without the WAP11 were almost identical. However, when I started streaming some really low bandwidth music (about

128Kbit/sec), the thruput was cut in half or worse. Actually, it seemed erratic and I spent some time trying to figure out the cause. My guess(tm) is that without flow control, I was getting some packet timing backoff variability. It may also have been the varying effects of different 802.11b modulation modes at different speeds. Dunno.

I don't know any AP's that have such a hardware switch. I'll assume you mean a software or configuration switch. In all routers that I've seen, the choices are roughly: 802.11b only 802.11g only 802.11b+g mixed. Is this the switch or did I miss something between the lines?

I beg to differ. With the typical rubber ducky 2dBi ommi and -2dBi PIFA laptop antenna, you're correct. 2.4GHz will give better range than 5.7GHz, all other things being equal. However, if you replace the junk antennas with external antennas OF THE SAME PHYSICAL SIZE, the 5.7Ghz antenna will have about twice the gain (3dB). If both ends have such an antenna, the 6dB total gain will yield twice the range at

5.7GHz.

As for building penetration, that's a crap shoot. 5.7Ghz will go through holes and slots much easier and better than 2.4Ghz. If the building looks like swiss cheeze, 5.7Ghz will work better. However, if the building is a solid wall of something (doesn't matter the material, just no holes are allowed), 2.4Ghz will penetrate better.

3com wireless antenna guide:

See Table 1 and note that some materials have higher attenuation at

2.4Ghz. My guess(tm) is that it's not the material, but the wide variations found in measurement methods and in "holes".

Compared for what? Max thruput? For *WIRED* routers see:

For *WIRELESS* routers:

There are not results for 802.11b impairment testing (yet).

Note that almost all the routers listed are 802.11n routers. The only

802.11g router on the list is the WRT54G2, which only did 13.6Mbits/sec or about 60% of theoretical maximum. Reality has a way of catching up with hype.
Reply to
Jeff Liebermann

I'll give you that, because he wavelength of the 5GHz signal is smaller than the 2.4GHz signal, so you may have a whole wavelength in antenna size instead of half a wave. Also, vendors are turning up the transmit power on the 5GHz channels to compensate.

That's the problem--in typical residential construction, the drywall is the main barrier, and that's a problem for the shorter wavelengths.

Steve

Reply to
Steve Fenwick

Where does it say "hardware"?

Trendnet TEW-510APB, 802.11a/b/g, with B/G mode usualy turned off.

Reply to
Axel Hammerschmidt

It doesn't. Your description was vague enough that I had to ask whether this mysterious "manual switch" is a real switch or a virtual software switch. Could I trouble you to identify this "manual switch"?

Perhaps I'm a bit dense this morning, but I don't see anything on that page that looks like an 802.11b protection on/off switch. The only item which seems involved is "Allow 54Mbs Stations Only" which seems rather vague. Does it turn off 5.7GHz? Does it disable speeds above

54Mbits/sec? Does it lock the router to 54Mbits/sec as the only speed? Got a clue what it means? It's kinda academic as you have it disabled.

So, where is the disable 802.11b compatibility mode "manual switch" you previously mentioned? Did I miss something (possibly on a different web page)?

Reply to
Jeff Liebermann

What's your question got to do with 802.11g 802.11b protection?

The switch on the fourth-, third- and second bottom line turn off

802.11g 802.11b protection, until 802.11b traffic is detected. Depending on how the switch on fifth bottom line is set. The choice is between different methods.

Below is another example, from the (CD-) manual for my ASUS WL-330gE multi-function pocket router:

"Wireless Mode

This field indicates the 802.11g interface mode. Selecting "Auto" allows

802.11g and 802.11b clients to connect to the ASUS 802.11g AP. Selecting "54g Only" maximizes performance, but prevents 802.11b clients from connecting to the ASUS 802.11g AP. If "54g Protection" is checked, GMode protection of 11g traffic is enabled automatically in the presence of 11b traffic."

As I read the last part, considering that 802.11b protection on 802.1g devices is required - see earlier posts, it means that "54g Protection" will be on all the time when not checked.

Can't you read AYBABTU?

Here's another one, from my routers firewall.

"If checked, the undefined item in IP Address list is allowed to access Internet;Unchecked means reverse."

Reply to
Axel Hammerschmidt

Why does asking you a question usually result in another question instead of an answer? I guess I'm doing the same thing. Do like I say, not like I do.

I was assuming that your mysterious "manual switch" was the "Allow

54Mbps Stations Only" setting (which you incidentally have disabled). The wording is a bit unusual and I thought that since you have the unit, have done extensive testing, and have possibly RTFMed, you might know what it really means.

However, I did make a bad guess. 802.11b compatibility is only an issue at 2.4Ghz, not 5.7GHz, so the setting does kind sorta make sense if "54Mbps stations only" really means "use 802.11g only".

I don't think so. Most of that is CTS/RTS "protection", not 802.11b protection. It protects against collisions caused by hidden nodes (where multiple clients can't hear each other and therefore tend to transmit simultaneously). At least that's what I think it means.

Ok. Now I see. Your "manual switch" is really a setting which completely disables 802.11b connectivity. Different router, different names, different description, but probably the same thing.

However, the original problem was over how an access point actually handles mixed 802.11b and 802.11g clients. I don't see anything that affects this except to totally ignore 802.11b clients. (Don't assume that I know how your chipset does this. I'm a few years out of date).

This is the first time I've ever heard it refered to as "54g protection". However, I must admit that it sounds better than "Disable 802.11b mode".

In the distant past, I used to work for Standard Communication in Wilmington CA. We sold VHF/UHF marine and land mobile radios. I still have many of the manuals that were first written in Japanese, and then translated into what they considered to be English. SCC hired a full time technical interpreter to clean up the mess. For the few that I helped re-write, it was easier to start from scratch. Direct translations of idioms were the major problem. I don't think I would do any better translating my English into Japanese.

Yech, but typical. A friends son is currently doing classes in Chinese at Humboldt State. He now talks and writes like that. There seems to be a cultural thing against being specific or fortifying a statement with a redundant statement. I do it conciously by avoiding the word "it" unless it's clearly defined (or doesn't matter). That's apparently not the case in Chinese, where brevity seems to be desireable. It also reminds me of some student papers.

I'll do the jperf benchmark test for 802.11b compatibilty sometime this week, after I drag home an 802.11b only client radio (WAP11). I'm curious how different routers respond.

Reply to
Jeff Liebermann

The number of routers involved may be interesting as well.

"There are three ways an AP will go into protection:

- An 11b device associates to the AP.

- An 11b AP on the same channel can be heard by the AP

- The AP hears an 11g AP that is in protection because of an 11b device associated.

The last statement is the hardest to explain, so here is further explanation:

11g AP goes into protection because of an 11b device associated. That AP will set the bits in the beacon as follows:

Non-ERP (11b) Present: 1 = Yes Use Protection: 1 = Yes

An 11g AP that hears that will go into protection and have the following Beacon bits:

Non-ERP (11b) Present: 0 = No Use Protection: 1 = Yes

Now, lets say another 11g AP heard the one above. It will not go into protection because the one it heard does not have the Non-ERP Present bit on. "

Reply to
LR

I'm afraid you're wrong.

Hidden node protection is set by twiddling the RTS/CTS Threshold knob. You'll notice there are two such knobs on the screenshot, one for when the radio is on the 5GHz band - the top pane - and one for when the radio is on the 2.4GHz band. There's only the one radio in this model AP. And you don't see the 802.11b protection switches in the top pane, because it's not needed for the 5GHz band.

Not all 802.11g AP's have 802.11b protection switches. My D-Link DWL-G700AP and DWL-G730 pocket router do not, f.ex. It's an ekstra feature. However, they do have an RTS/CTS Threshold knobs.

Reply to
Axel Hammerschmidt

Probably, but I'll never admit it. So, why is it called "CTS/RTS Protection Mode" and "CTS/RTS Protection Rate"? CTS/RTS flow control has nothing to do with "802.11b protection" (which really should be "802.11b legacy compatibility". What does the "help" button say about those two settings?

I beg to differ somewhat. Most routers have three adjustments when dealing with flow control. 1. CTS protection mode on/off 2. Fragmentation threshold 3. RTS threshold The first turns the mechanism on and off. The 2nd determines the maximum packet size. Smaller packets have a higher probability of actually getting delivers. The 3rd determines the minimum packet size that will *NOT* require flow control. That's because the CTS/RTS overhead for dealing with small packets has a huge effect on throughput. Small packets don't get or need flow control.

I can sorta see some vestige of the common terminology in the Trendnet wireless setup page. If my astute guess is correct, they've either split the CTS/RTS settings into two places, added some new settings to deal with dual band operation, or totally mislabeled what you assume is control over 802.11b compatibility. I think I could make a better guess if I knew the options available in all the settings. If my guess is correct, there's absolutely no setting on that page involving

802.11b compatibility and that all the CTS/RTS flow control settings actually involve flow control.

Incidentally, on my wish list of router features is the ability to set parameters for different bands on seperate pages. Mixing them together, on one page, like most router manufacturers like to do, is confusing.

Is there an online emulator for Trendnet?

I just checked a few Linksys routers at:

for how they handle 802.11b compatibility. Most have no fine control over compatibility, giving only "off" or "auto" as choices. Those with 802.11n features add "N only" to the confusion. Not much extra in the CTS/RTS flow department.

I like simple theories. If it says CTS/RTS in the setting, it must be CTS/RTS and not something else. Check the Trendnet help button.

Reply to
Jeff Liebermann

The number of interruptions and crisis problems that arrived yesterday may also be a problem. I have to do this at home because there are too many active wireless networks near my office. With a glass wall, I can't easily block them (tin foil curtains?). Benchmarking done at my office has been eratic. At home, I have the absorptive shielding provided by a wall of redwood and fir trees. They absorb 2.4Ghz like a sponge. However, I've been getting home at midnight lately and the last thing I need is homework. Maybe next weekend.

So, now I have to drag out the protocol sniffer (Wireshark on Linux) to verify the protection mode. Without some way to determine if it's in 802.11b compatibility mode or not, this is not going to be easy (or consistent).

I was planning to just repeat a previous test, adding a few additional cases:

  1. 802.11g benchmark with no 802.11b device present. This should be the fastest.
  2. 802.11g benchmark with 802.11b client device spewing beacons, but not associated or moving traffic.
  3. 802.11g benchmark with 802.11b client device associated, but not moving any traffic.
  4. 802.11g benchmark with 802.11b client device associated and moving low levels of traffic (i.e. streaming audio).
  5. 802.11g benchmark with 802.11b client device associated and moving as much traffic as possible (i.e. file download).
  6. A few of the above with CTS/RTS flow control enabled and set for fairly small packets (512 bytes). I've always been suspicious that CTS/RTS flow control is really only effective in homogenous systems and will screw up throughput on systems with a wide mix of speeds. Also, some tweaks such as CTS only, if available. (This one is a real time burner).
  7. Try most of the above with the AP speed locked to 54Mbits/sec to see what breaks.

Nice find. I never thought to search there.

I got somewhat lost in the timing calculations as presented. I noted that he neglected any mention of the 1Mbit/sec beacons that are required for 802.11b compatibility. These are large airtime hogs.

I have a copy here, somewhere, maybe... He also assumes that both

802.11b and g will be running at the maximum possible speeds in mixed mode. What I've seen is that the AP almost immediately shoves the 802.11b connection to the slowest possible speed in order to reduce errors.
Reply to
Jeff Liebermann

Go back in the thread, to my first follow-up to Steve Fenwick's post. It has

Message-ID:

and take a look at Gast's Fig 14.1, Basic overview of protection mechanism. And the text above. That should explain it.

(Don't worry about the language in the Google stuff. For some reason it defaulted to Danish).

Reply to
Axel Hammerschmidt

Methinks not so basic. I found your link for it at:

Well, I have that book but none of the 802.11b protection stuff is in there. However, I can see why. There are apparently multiple additions of the book. The version that Google books finds is at:

Chapter 14 has nothing on 802.11b protection (unless I missed something).

Customers cometh... more later.

Reply to
Jeff Liebermann

You've got the first edition. That edition covers 802.11b and 802.11a. Not 802.11g - and hence not the bit about 802.11g 802.11b protection.

The Google Books link (the Tinyurl) is to the 2nd edition's figure 14-1 (not 14.1). I did a Google search using the string: "Basic overview of protection mechanism" - with the quotes.

Reply to
Axel Hammerschmidt

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.