a question about channels?

I have a ware house that has 6 AP's. They are broadcasting on several different bands.... 1, 6 and 11 are all accounted for. Would it be better to have them all on the same band?

Reply to
200mg
Loading thread data ...

No, you have to separate them so that AP's are on no-overlapping channels.

For instance, an AP on channel 1 should only be able to see AP's on channels 6 and 11. Likewise for an ap on channel 6, or channel 11.

If you're getting a lot of interference, I'm working with a company that has built a wireless networking system that works on a single channel, and has zero-latency between access points.

If interested, give me a call.

Chris

- - - - - - - - - - - - - - - - - - - - Christopher M. Hutchison, CEO NetSteady Communications, Ltd.

Phone: 614-255-5575 Mobile: 614-853-0091 Skype: wifi_chris

formatting link

200mg wrote:
Reply to
NetSteady

You have a problem. Each channel number is 5MHz from the next channel. However, the 802.11 signal occupies about 22Mhz or about 5 channels wide. Therefore, to prevent interference, the only non-overlapping channels are 1, 6, and 11. If you pick a channel in between these, you'll get interference from BOTH channels on either side. The trick is to limit the radiation pattern from each xmitter using fancy antenna patterns, allowing you to re-use channels in different parts of the warehouse. Otherwise, just turn down the tx power until the interference is reduced. It will never go away completely, but lower power does help.

See the hexagonal channel pattern mentioned in the Intel Hotspot Guide in section 5.2.3.1. The link on the Intel web pile went away and I can't find the replacement. For now, here's my copy:

I suggest you skim the whole thing as there's quite a bit of really useful info buried in it.

Reply to
Jeff Liebermann

the situation I have now is that one AP is reporting 87% signal, then if you walk literally 3 ft in one direction it drops to zero. The 2 closest AP's were xmitting on chan's 1 and 11, the ap in question was also xmitting on 11, so i changed it to 6, with the same results. I also tried 11 again with a lower power setting and still had the same results.

There is apparently alot of power conduit and coming out the top of the office I am trying to extend this signal to. I am also going top try some antenna with a higher db, I just find it odd that it goes from close to 90 to 0 in such a short distance.

Jeff Liebermann wrote:

Reply to
200mg

Is the AP reporting the signal strength (i.e. using SNMP or some kind of managment tools) or is the client you're using reporting the 87% signal?

Also, I don't think my walking in any direction will have an effect on your signal strength. However, if you do the walking, you might see a change.

Wireless doesn't work like that. It will fade away, not die suddenly. The only way I can think of creating such a fast drop off is to have substantial interference in the area on the same or adjacent channels. As you get closer to the interference source, the desired signal goes away. Your system might also be switching to a different access point, which appears as a loss of signal.

Fine. Try turning OFF all but one access point and see how the client signal strength behaves. I think you'll get a really good clue as to how much interference you're dealing with by testing just one access point at a time, with the others turn off (that's powered off, not the ethernet cable unplugged).

I'll pass. You're not supplying any hardware or topology details and concentrating on the channel layout. While this is important, I suspect you have bigger problems. I can't suggest much beyond basic troubleshooting (by elimination) unless I know the hardware, topology, layout, and setup.

Reply to
Jeff Liebermann

On Fri, 01 Dec 2006 05:17:29 GMT, Jeff Liebermann wrote in :

Perhaps we should add an item to "Top Ten Fixes to Wi-Fi Problems":

  1. It's not what you think -- something else is going on.

Or moving from an open doorway down which the signal has been traveling into a room with metal in the walls.

Multipath interference?

Never fails to amaze my how many people take the scattergun approach to troubleshooting, instead of the far more productive step-by-step approach.

What?! Not willing to waste time in wild guesswork?! :)

Reply to
John Navas

Thanks for your replies.

About the topology:

There are 6 AP's spurring from 2 linksys unmanaged switches. The one in question exists about 80 feet from the office I am trying to get this signal to extend to.

I am using a symbol scanner with pocket pc2003 on it and some symbol software on these scanners to test for signal.

Now in my warehouse in Cincy, the signal behaves just as you mention, it fades away the further you walk from an AP. I can tell this by the results given to me by the same means I am testing in my Baltimore warehouse. In Baltimore, I'll be 10 feet in front of this office, check for signal, it will say %85, i will take literally 3 steps toward the office and it drops to zero.

It's not re-associating with another ap because while it's up there are

3 visible AP's, then when you get closer by approx. 3 feet it loses not only the one it's connected to but the other two as well.

There's not necessarily anything different between the two warehouses. I cant see anything on the ceiling that would interfere with the signal

TIA

I'm us> On Fri, 01 Dec 2006 05:17:29 GMT, Jeff Liebermann

Reply to
200mg

"200mg" hath wroth:

I kinda wanted the maker and model number of the devices (access points, antennas, clients, hand-helds, etc). The reason is that different hardware and devices behave differently. For example, a common way to do a warehouse is to use leaky coax. I know how that behaves. Another common method is a very directional antenna shooting down the aisles of the warehouse. That will cause a rapid fade as you turn a corner or move behind a stack of absorbent or reflective merchandise.

Basically, you cannot "flood" a warehouse with RF any more than you can do the same with light. If you put overhead lights on the ceiling, and then block it with merchandise, you're going to have dark zones. Same with 2.4Ghz but worse as you will have diffraction, refraction, reflection, and absorption effects.

What I was hinting is that the channel juggling is important but that something else is apparently happening. Wireless does not magically fade away that quickly. I've done troubleshooting jobs like this before including one big produce storage warehouse from hell. Their problem was similar to yours. They had too many access points (thanks to the salesman). Whenever there was a coverage problem, the sales guy just sold them another access point. 200,000 sq ft and they had something like 20 access points when I arrived. Nothing worked because of mutual interference. I reduced it down to about 11 access points, with very directional antennas to allow me to re-use channels and to get great coverage of the aisles near the ground level. It didn't work too well when the warehouse was empty or near the roof, but that eventually turned into a useless requirement. Management didn't want workers punching buttons on the radios while 20ft in the air balanced on the edge of a fork lift. Again, note that the antenna selection was more important than the interference reduction. When I was done, only one of the AP's had omni antennas.

In order to get it right, I had to map the coverage area of each access point individually. It was a tedious pain and I hate doing it. I had a large sketch of the warehouse floor space. I walked around with a laptop running Netstumbler, picked a relatively strong signal level, and drew the approximate contours. Accuracy was awful, but I got the general coverage area. By combining the plots for each access point, I was able to demonstrate that some were redundant, some were useless, and some were mis-located. Management was impressed, especially when I told them that they could use the AP's I ripped out in the office area. There are commercial products that do this and/or attempt to predict coverage.

formatting link

Ok, so you've seen how it's suppose to function. The question is "what's the difference". I don't think it's channel selection, although it's possible.

The only way to troubleshoot a complex system is to rip it apart and build it up from scratch. As I suggested, turn OFF (as in power off) all but one access point and see how it plays. My guess is that it will work just fine in the immediate area. Try to get a feel for how much RELIABLE coverage it has with just one AP. Then start turning on the others, one at a time. Start with one that's on the same channel which should give the worst case results. I usually monitor the traffic with an SNMP managed switch but you indicated that the Linksys switch was not managed. Oh well. Use file transfer throughput as an indication of reliability, not signal strength. A highly reflective environment will give a great signal strength, but crappy thruput.

Ok, that eliminates AP hopping as a problem. Symbol AP's have a feature called "automatic channel selection". Hopefully, you're NOT using this feature as it has caused me nothing but grief when I tried it. If things seem to work just fine one day, but truly awful the next, look for this "feature".

Also, I have the habit of disabling automatic speed selection. 802.11g will tend to go for the fastest connection. It will continuously try to go as fast as it can possibly go. This is great for bloated applications moving massive files around, but a real PITA when dealing with small bursts of data. I usually fix the RF speed at 12Mbits/sec which is more than enough for most interactive apps. A side benefit is that the usable range tends to improve.

Keep going? Which Symbol scanner? 802.11b or 802.11g? Omni antennas on the AP's? Lots of ways of causing problems.

Incidentally, don't ignore the possible sources of interference. The same warehouse I previously mentioned had a 2.4GHz cordless phone base located a few feet from one of the AP's. There was plenty of interference when the phone was active. Nobody connected the interaction because different people used the phone and the scanners.

Reply to
Jeff Liebermann

John Navas hath wroth:

That which is most obviously correct, beyond all need of checking, is usually the problem.

Assumption, the mother of all screwups.

Yesterday afternoon, a customer calls to announce that their server is dead. No screen, no boot, but the drives spin. I arrive and determine that it's dead. I drag the box (about 50lbs) to my palatial office, plug it in, and it works. Tinker for an hour and don't find anything wrong. Drag it back, plug it in, and it's dead. I eventually find a shorted keyboard. Well, at least I got some good exercise.

It's not that abrupt, even in a doorway. 2.4Ghz tends to refract from apertures. However, if there's an interference source through the doorway, that's very possible.

No. Frequency selective interference comes and goes as one moves around. Also, the diversity reception in the access point does a good job of reducing any such effects.

Add me to the list. I tend to do what easiest first, even if I know it's well down the list of probable culprits. I like to reduce the laundry list by eliminating the easy stuff first. To the casual observer, this resembles a disorganized effort. Troubleshooting is not something the come naturally and certainly not without a good understanding of how such wireless systems operate. I would not expect an intelligent beginner to be very good at it. That's why I often suggest troubleshooting by replacement, which both easy to accomplish and understand. When we get into complex wireless systems that include everything from propagation to topology, everything starts to interact with everything else. That's why I suggested ripping it apart and build it up in pieces. Problems are far more obvious when the systems are simplified.

Knowing when to give up is the secret of engineering. I guess it also applies to troubleshooting wireless networks. Juggling channels might have an effect, but my sense of smell tells me that something else is happening.

Reply to
Jeff Liebermann

Jeff Liebermann hath wroth:

One more:

formatting link

Reply to
Jeff Liebermann

thanks for all of your help jeff. I'll try everything you mentioned.

we don't stack to the ceiling, there is probably 18' clearance from our highest point to the ceiling

180,000 sq. feet warehouse 22' ceiling 6 Symbol 4131 AP's c> John Navas hath wroth:
Reply to
200mg

"200mg" hath wroth:

That actually makes it worse. With a 22ft ceiling, and 18ft of clearance, every access point can probably see every other access point. Same with the scanners seeing all the access points. With the stock omni antennas, you certainly are going to have an interference problem with 6 radios on 3 sets of channels.

However, if your access points are near the ceiling, it's possible to illuminate a large area with only 4 access points (near each corner). However, you can't use the omni antennas that come stock with the

4131. You'll get far too many reflections off the ceiling and walls. Some type of low gain panel antenna will work. Circularly polorized panel hung on the ceiling with about 6dBi gain and about 60 degree beamwidth. I'll work out the geometry if you want (gotta run right now).

Is that 425ft x 425ft? With 4 access points and 200ft cells, that a maximum range of about 100ft for each access point. That's a bit far, but not impossible. More gain would help, but unless the warehouse is rectangular, your original 6 access points will probably be better. With 6, the cell size shrinks to: 180,000 / 6 = 30,000 sqft/cell Assuming a square cell: cell radius = 1/2 * sqrt(30,000) = 87ft Better.

Retch. The 4131 is 802.11b only. 802.11b doesn't work very well in a highly reflective enironment. They also belch 100mw (+20dBm) which might be a bit much if you're worried about mutual interference. See if lowering the xmit power helps.

The 9060G is also 802.11b only and also belches 100mw (+20dBm). I would like to suggest you dump everything and switch to 802.11g, but that might be a bit premature. Methinks you should find out what is causing the problem first. I think turning off the other AP's and looking for other possible sources of interfernce would be a good start. There has to be a reason why the scanners act weird in this warehouse, but not in the other.

Incidentally, see what I mean about supplying numbers?

Reply to
Jeff Liebermann

On Fri, 01 Dec 2006 10:45:24 -0800, Jeff Liebermann wrote in :

That wasn't my point. I was talking about changing multiple things at one time. With (say) 8 possible causes, change all 8 items. When the problem isn't solved, nothing has been gained. Better to take them one at a time, in whatever order.

And here I thought the secret was knowing when the design is done, instead of endlessly trying to improve it.

See my #11 above. ;)

Reply to
John Navas

No engineer working on a project ever really knows when it's done. The project has to be ripped out of their hands, while they kick and scream that if they only had another few days, they could have solved all the bugs and problems that were ignored in the past 12 months.

Instead, the proper way to finish a project is to scream that they're fed up, throw a few things around the lab, and announce that they're gone for a much needed vacation. This method minimizes the number of casualties typically found under the workbenches near project end, and insures that there's a clear path for the engineers in manufacturing engineering change everything and redesign the project from scratch.

I find it odd that the colleges and skools fail to teach prospective engineers how to give up in disgust or how to gracelessly end a project. These are skills that all practicing engineers must develop in order to survive more than a few projects. I suspect the lack of this important skill is why the early burnout rate is so high in engineering.

Reply to
Jeff Liebermann

On Tue, 05 Dec 2006 04:43:36 GMT, Jeff Liebermann wrote in :

I find it equally odd that the colleges and skools fail to teach prospective engineers how to calculate the value of diminishing returns. ;)

Reply to
John Navas

John Navas hath wroth:

Huh? My 2nd favorite kollege klass was "Engineering Economics"[1] which included time value of money, make buy lease rent decision, taxes, depreciation, estimating, and just about all the financial aspects of engineering. I recall that diminishing returns was mentioned. I still have the text book and occasionally use it.

Later on, "The Mythical Man Month" and similar books:

detailed the problem. I do take issue with the alleged exponential nature of diminishing returns on effort and find the linear model to be more appropriate for projects. If one person can do it in 1 day, two people will take 2 days, 3 people will take 3 days, and so on. After about 5 people, a non-productive manager will need to be added, which will approximately double the completion time.

Fortunately, today's skools do teach these things but by example. I was gratified to see class projects and group efforts for designing and building real devices. This is quite different from the "thou shalt not collaborate" dictum that was popular while I was in skool. With these group efforts, I can only assume that the intelligent student will have some experience in group dynamics and their effects on performance, such as diminishing returns. One can also assume that the student has read at least one of the Scott Adams "Dilbert" series of management books, and discovered that much it is factual. Whether students are able to calculate the exact value of diminishing returns is debatable, but one has to start somewhere.

[1] My favorite class was tractor driving and mechanics.
Reply to
Jeff Liebermann

On Tue, 05 Dec 2006 11:55:19 -0800, Jeff Liebermann wrote in :

My favorite class was sailing. OK, not a real class. Damn hard to concentrate in Physics class on a warm spring afternoon with sailboats easily seen through the window. :)

Reply to
John Navas

You went to Purdue?

Reply to
Kurt Ullman

Kurt Ullman hath wroth:

I just knew someone would ask. Nope. It was Cal Poly, Pomona.

It was primarily an agricultural college at the time, but had a decent engineering skool. It was also the only college that would accept me after getting thrown out of San Fernando Valley State in Northridge CA, which was also the only college that would accept me after getting thrown out of Santa Monica City College. It was easy to recognize the engineers and the ABM (agricultural business management) students. The aggies wore suits and carried brief cases. The engineers wore jeans, boots, and 10" slide rules. At the time, I was majoring in draft dodging and loan sharking. In order to stay in skool, I had to work part time to pay the bills. That meant that I had a limited selection of classes available to accommodate my work schedule. I needed about

15 quarter units to avoid the draft and agricultural technology classes were considered marginally acceptable. So, I took tractor driving and later tractor mechanics. I don't hold the record for maximum damage in a single quarter, but I came close, with about 50ft of chain link fencing, one water tank, and almost burying a D6 track layer, to my credit. These classes were especially useful for working on Cal Poly's Rose Float entries, which I did even after graduation.

I was graduated in 1971, just in time to see the space program collapse along with all the nifty jobs. So, I ended up installing radios in cement mixers and fixing tractors for a while. Again, the two classes were invaluable.

Engineering economics was also not neglected in my career. Having helped build the college FM radio station with 100% stolen materials, I was well versed in the practical economics of project engineering. The experience proved invaluable in providing the background necessary to calculate the cost/benifit/risk/bail/jailtime ratios involved in borrowing test equipment, stealing parts from other divisions, burglarizing incoming inspection, and internal espionage. The motto of Cal Poly is "Learn by Doing" and I was more into over-doing.

Incidentally, I rode in the float shown that year. If you look above me at the large hinge point, you'll see a hydraulic hose. That hose leaked and gave me a hot oil shower for the entire length of the Rose Parade.

Reply to
Jeff Liebermann

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.