site survey apps?

[guise]

anyone have URL to free or low cost site survey apps that work with any wifi NIC? Some or even most of the ones that are listed on this site work with only specific nics or AP. I would reaaly like to perform a reasonable site survey using my laptop with my graphical floorplan. i will be replacing all current workstations with wifi laptops and anticpate also replacing our current phone switch with a nortel VOIP solution. The end result will be a complete wireless LAN and VOIP. www.wdtc.net/wdtc.jpg

i have about 25 users that maintain a minimum 1Mbps with all immediate applications. it seems i will need to have around 5 AP with directional antenas to cover the area and not create unecessary overlap of channels. i would like to learn how to do this onw my own without having to outsource the work.

[davidjwieland]

I really like using this...once you get used to it, it can become really really handy.

www.netstumbler.com

Best,

[guise]

Without knowing specific site survey data. Could anyone offer their opinion regarding the two different AP location setups shown below? Why would be better than the other (wdtc_1.jpg or WDTC_2.jpg)

http://www.wdtc.net/wdtc_1.jpg

http://www.wdtc.net/wdtc_2.jpg


Also, as i mentioned above the main reason i would like to have 5 or more access points in such a small space is to give more bandwidth to each user by distributing as many users as possible across as many AP as possible and keeping each user within the AP 11Mbps opperating range. Each user currently uses at least 1Mps total for all their network applications sometimes a little more. Also, I am keeping in mind that in the near future (6 -12 months) i would like to extend our WLAN to support Wireless VOIP.

Some of my concerns and question are thus so far:

1. Do i need to be concerned about RF interferance from one AP area on one channel into an area of another AP area opperating on a different channel.

2. Is the amount of distance that I have shown ~55 feet enough distanance to prevent RF interferance from AP opperating on same channels with typical directional antenas?

3. If answer to #1 is NO and #2 is yes. Is it very difficult to configure the AP to behave as needed in the described/shown environment/setup (assuming there are not any other unaccounted for interferances).

sorry for all the questions but your input is appreciated. it seems the more i read the more questions i get as i learn more about my new project.

thank you in advance,

scott healey

[davidjwieland]

Scenario 2 is far more favorable, except that the hard-wiring will be a touch more complicated...well worth the added performance though. Also, I would use channels, respectively, 8, 4, 11, 6, 1 in that order from left to right or right to left. That will keep your interference to a minimum. This won't be a difficult setup, so I wouldn't worry too much.

All my best,

[Uncle Ken]

My suggestion is to use 802.11a. Better bandwidth, more non-overlapping channels, better signal containment, less interference from cordless phones, microwave ovens, other people's 802.11b networks, etc.

[guise]

yes after further research i am tending to agree the #2 makes for better layout. The hardwiring is virtually in place due to existing LAN.

What would be the difference in using channels: 8, 4, 11, 6, 1 vs. 11, 6, 1, 11, 6? I thought it is recomended in site surveys to establish the boundaries of access point coverage to prevent more than three access points from mutually overlapping, unless certain areas use multiple channels in a single area for greater throughput and to keep a 5 channel sepreation between overlapping channels in order to to prevent inter-access point interference. Doesnt this allow access points to have overlapping coverage areas with full throughput, provided the radio channels differ by at least five?

Uncle Ken,
yes .11a seems more appealing. I'll need to read up on .11a and will strongly consider it as a likely implementation.

Thanks for comments and suggestions.

I have another layout in mind i will post in a moment this would be similiar to #2 but locate the AP more in the middle and utilize a Dipole antenna.

[guise]

here is option #3 layout with AP more in middle of each cell utilizing a Dipole antenna. Do you think this could be preferable over #2 (located in above post)?

http://www.wdtc.net/wdtc_3.jpg

I think wdtc_3.jpg operating at .11a maybe be my latest tendency.

[davidjwieland]

802.11a is preferential for functionality, not for price though...yet. And the reason you want several different channels is that, from your drawings it looks as though the space is not that large. If you have three APs all on channel 11, you WILL have interference and it will inhibit your throughput and performance greatly. if you use 5 totally different channels, with all adjacent cells' channels spread relatively far apart, you will not run into this problem.

Best,

[Uncle Ken]

I think you'll find that dipole antennas have more of an omnidirectional pattern. BTW, is this a single story structure, or are there neighbors above and/or below?

[guise]

ken, its a single story structure. no additional buildings within 50 yards. we are a transportation facility on a large lot. our shop is adjoining. our building but that is irrelevant. in an additional phase we will extend wlan into the shop ~ another 30,000 sq ft.
i'll look into the dispersal patterns for Dipole antennas. The dispersal pattern diagrams i've seen seem to server my placement in http://www.wdtc.net/wdtc_3.jpg

david, you'll seend on my floor plan that i am anticipating about 5 AP. If i choose to go .11a or .11b and from my understanding (which has a strong likelyhood of being incorrect) if my channels are arrange 11, 6, 1, 11, 6 all potentially overlapping cells have at 5 channels of seperation the nearest same channel cells is at least 40 feet distance from each other with 10 channels of speration with its signal direction not directed at each other. But i do see you method has no same channels.

(note: my possible ignorance) can you use channles 4 and 9 in .11b environment? are there any disadvantages?) I thought .11b had only 3 channels to utilize without suffering loss of throughput.

thanks for your patience

scott

[jimmyjam]

Just a quick thought, 802.11a is a risk because they want to use wireless VoIP eventually and you're banking on people making devices to run on it.

[davidjwieland]

Not following your logic...802.11a allows for more throughput, which is what you want in an office, ESPECIALLY if you're trying to run VOIP.

Best,

[guise]

hmm, we are looking to use the Nortel Business Call Manager 400 (BCM400) and Nortel AP 2220 the ap2220 supports both .11a/.11b As i skim thru the literature of the BCM400 i see it mentions something about Mobility- Symbol VoIP Wireless 802.11b Datavison and Netvision phone. What that means i dont know yet ill have to read over it some more. Since all workstations will be replaces with wifi laptops with softphones and the desk will also have VoIP wired phone on the desk using LAN. We will have mobility with VoIP at .11b speeds utilizing WVoIP as well as VoIP phones on the desk using wired infrastructure. (does that make sense?)

but it does seem posible the the wifi laptops and softphones might not be able to utilize .11a functionality. (shrugs) still more researching and we have a meeting scheduled with potential on july 8th. so this seems like a reasonable scenario to put forth.

[spiderbite]

It is going to take some doing to get 5 ap's to load balance in less than 6000 sq feet of space. The power would have to be way down, and that might not do it depending on wall construction.

Could you consider an oversubscription solution? Of the 25 people, how many really are hitting the network at once? By using a 3x oversubscription, you could accomodate almost 50 people at the throughput you require with 3 ap's.

Something to think about.

BTW, any reason to slow down the laptops on a desk when there is a 100Mbps CAT5 jack at the wall?

Wireless is handy, but dont kick yourself when you dont have to.

[davidjwieland]

Wire it when you can...your VOIP is going to severely bog your wireless network down. And no one wants crappy voice...you can do without streaming internet service, but not without streaming voice.

Best,

[guise]

spider, thats a valid point about the laptops. But, heres the deal. We have old p3 500mhz systems, that are just ****ing on performance for the apps we use. I currently am running all the apps on my laptop P4 2.4ghz Mobilty with 512mb and fast 40gig drive and on a wireless connection dedicated for my sole use faster than any one of our desktops. So, instead of buying new worstations (which initially we planned) why not increase performance for all users, give them mobility, give the ability to take their "desk/pc/phone" home with them ( squeeze out more productivty and give sense of ownership and freedom having a sweet system. I have the ability to do EVERYTHING i do in office on my laptop from home via broadband (manager servers, configure sql server, apply scripts to db, monitor web, email connections and apps i.e. anything). why not give the rest of the office the same ability with a simple focus of direction to upgrades we would already be doing without a long term vision of true office mobility. Once we add VoIP and the abilty to work completely work out of the office %100 with an increase in functionality some decrease in technology (i.e. 100mbps wired vs wlan) but yet a total gain in performance. the answer seems obvious. if the money is there and we plan on spending it. why not complete an orchestrated vision and come out with something that everyone looks at and says !@#%% yeah!
(sorry i just ranted that and no time to proof. hope it makes sense) so to make long answer a bit longer. we will forego the 100mbps jack because 1. the standard ip phone will be using it at most deskss 2. who the hell like to plug their laptop in when they have wifi and the difference in performace in most instances is negligible) some times at work i wish my laptop had more bandwidth. but really i like it sitting clean on my deskwith only a powercord and mouse cord. (on a seperate note: i wish my inspiron 8200 had builtin bluetooth and i could find a nice bluetooth mouse to accomodate it AND when we gonna get wireless powersources that plug into a wall outlet and transmit power to laptops;))

Also, I was thinking (and i may be very well be thinking wrong) that each AP with a directional antenna would service a cell 20 feet wide and 58 feet long. between using the directional antennas and (tweaking power if needed. i dont at this time see it necessary) spacing same channel cells at least 2 cells apart (about 40 feet) or no same cell channels if decide to go .11a. Currently each users currently maintains a constant 1Mbps (minimum) connection averaging 1.5Mbps (with current apps no VoIP)

could you explain more detail how an oversubscription soltion would work in more detail. by my math i dont think it would work.

regarding VoIP slowing down. im not to sure about it. jly 8th ill be visiting a site that has a similiar implementation. Voice doesnt require much at used with good QoS solution your voice wont suffer regardless data traffic. notto mention 90% of voice in office will be wired. Wireless voice is just an option when people are moving around the office away from normal working area and at home or on the road or offsite at corp locations.

great feedback guys thank you! you keep giving me something to think about and requiring me to think in more depth about my preconceived notions.

[spiderbite]

Where did you come from? I wish everyone who had the ability to spend the money had your enthusiasm for this stuff. Make my job a hellofalot easier.


Since you have provided a valid argument in spite of conventional wisdom, I will agree to your point and no further discussion is necessary.

First thing is first- when you get around to doing the survey you will find that the coverage patterns are probably not going to behave as expected. Without some serious attenuation from building construction i.e. Brick walls, I just dont see it happening.

So, if your office or space is mostly sheetrock construction and there is not much else absorbing the signal (like floor to ceiling reams of paper) you will probably find that the coverage pattern is too wide. More to follow

[spiderbite]

Since directional antennas have more gain you may find out that the coverage will be too much for your purposes. I assume you are trying this in a sheetrocked environment with not much else in the way. A balance has to be struck between antenna selection, power setting, coverage, and throughput that will perform as advertised (or as hoped for). Too many AP's in a room that cant load balance is just tripping on itself. If you stay associated to the same AP all day and never roam eventually one AP will get hammered and the others never get used.

So - when you do the survey play with different antennas, and different power settings. If the cells are small enough you wont have a problem with channel allocation:D .

Im getting there...

[ccrum]

Interesting thread. I agree with spider about the coverage patterns. Indoor, you can forget what the patterns look like that were generated out on a range. These are far-field patterns only and if you are within 50 feet of these antennas at these frequency, it is impossible to tell what the patterns will look like. Not only that, but things like walls, desks, ceilings, and any other surface will reflect signals causing large amounts of multipath meaning every "site" will be seen everywhere in this size office. There really are only 3 non-overlapping channels to use to completely eliminate interference problems. If you are careful about placement, you could probably get away with doing a 2 channel spacing and still not have too much of an interference problem. After all this is spread spectrum.

I would stay away from using Omni antennas if you can. I would also mount antennas that cover different "zones" back to back (about 1.5 feet apart to avoid mutual coupling) instead of on opposite sides of the room. This gives a bit more isolation between antennas. Use as narrow beam antennas as you can get (something less than 30 degrees if possible, and target the center of the "zone". This way you are setting up a sort of "Cell site" in the middle of the office with "Sectors". There is a reason the cell and PCS guys do it this way. It works.

[spiderbite]

Back to oversubscription- A Cisco 1200 ap will on avg. yield 5.5Mbps of actual, useable, throuhghput. Based on your requirements of 1 Mbps per user, this ap will comfortably accomodate 5.5 users. Ok so 25 people need at least 5 ap's.

with me so far..? Ok now surely we can surmise that not every single one of those users will be striking the enter key or be talking at the same exact time all day....see where Im going..

So lets assume that only one out of three ( Just pulled it out of the air) will actually be somehow engaging the network at once- tada...3x oversubscription! This means that my one ap at 5.5Mbps could theoretically support 16.5 users. (5.5 x 3).

That means with only three AP's ( which would solve my problem of having too many) I could theoretically support 49.5 users. Or have lots of room for growth and or error!


So no 49 people arent on the network at once but since only one out of three might be...There you go.

Cell phone companies do this with your cell phone but it is more like 200x oversubscription. Which is why your phone doesnt work at the mall during Christmas!


Hows that?

[spiderbite]

Actually, If nothing else considered you could deploy this successfully with a bunch of differnt antennas.

My thing is to use the one that is the EASIEST to deploy that looks good. I do not like my installs looking junky.

If it were me, I would probably get my hands on some diverse microspheres which can be mounted above the ceiling grid along with the AP and nothing is visible below except two nylon screw heads on the outside of the tile. Technically, this would be an omni kinda thing, but my cells are going to be small and this is only 6000 sq.ft. Plus I get the added bonus of a diversity thing to help with any multipath I might do to myself, which I cant do with a patch. plus the gain is much like a ducky but the efficiency is much better.


Hell, I could have done it faster than it took to explain it:D

[guise]

spider, ccrum, david, ken and jimmy thx for your input comments, suggestions and ctitiques. i want to take a bit and digest what you guys just explained and do some more reading. i'll prolly come up with some "but what if'"'s or "how about"'s and "maybe if"'s. If nothing else I'll have a reasonable start when the Nortel sales/tech representatives come out to our site on july 8th and make an effort telling me what i should buy how to do. (i hope i dont drive them crazy! they might charge me double for being a pain in their arse!) I just dont want to be in a position that they come into the office and hook me up with some slock setup.


i really wish i knew how to use my personal Dell true moblie card in my laptop and the true moble AP to do some prelim site analysis. by toting them around the office and being able to sample, graph and map them to my floorplan. most of the apps cost some major $$, or dont work with my hardware, or are just simple utilies like netstumbler that i dont know how to make server my purpose.

[ccrum]

It's funny. After running an RF Engineering consulting firm for 6 years, you see the same things over and over. People don't want to spend the money to do it right the first time, but there always seems to be enough to come back and spend double to fix it later... after heads roll. The truth is, if you spend some money up front and buy the test equipement you need, or hire a qualified ENGINEERING firm, not an IT firm who has thrown up a couple of WLANS, then chances are you will have a network that will be robust, have the capacity you need, and not crumble at the first sign of interference. Sorry IT guys, I'm sure some of you know what you are doing, but I've fixed more installations done by "certified" people than I care to talk about.

I highly recommend the Berkeley Yellow Jacket (www.bvsystems.com). It's a little pricey at $3200, but it has a built in Spectrum Analyser, it will show you usage, paket error rates, received signal, correlation, delay spread, and signal quality for EVERY AP and on every channel in range. It even has a proximity detector so you can hunt down rouge APs. There is an add on package (BirdsEye) that will allow you to put in your floor plan and record data at locations for every AP. You can then downlowad the data to your computer and analyse it all there if you wish to see what signals are showing up where. I use it and wouldn't use anything else. For $3200 you get the receiver, a really nice case, and a Compaq 3955 IPaq. It all fits in your hand nicely. I really don't think it has an equal and the price isn't that bad if you want to do stuff right. I used to lug around a big Spectrum Analyzer and try to use Netstumbler at the same time. Trust me, it's worth it not to have to do that anymore.

[jimmyjam]

Originally posted by davidjwieland
Not following your logic...802.11a allows for more throughput, which is what you want in an office, ESPECIALLY if you're trying to run VOIP.

If you know anyone producing or talking about producing an 802.11a VoIP phone let me know please.

[davidjwieland]

Has them in the pipes from what I've heard, but nonetheless, there was some miscommunication here. I was under the assumption that all the phones, even the desktop ones, would be running over the wireless network (which apparently is not the case, and is good judgment on the the network designer's part). In that case, one would simply connect a desktop IP phone like the cisco 7960 to a wireless ethernet converter and that's how one would go through the wireless network.

Sorry for the confusion. Good luck!

[seadog]

guise

Since you are looking at a dual-mode AP (802.11a/b) I would recommend running your wireless voice clients on the "b" side of the network and your data clients on the "a" side. This separation should allow plenty of bandwidth for both sets of clients and your voice quality with the numbers you are talking about should be fine.

Also since the "a" side of your AP can support 54Mbs raw or ~27Mbs true throughput you can probably get by with only (3) APs, (3) by 54Mbs is a lot of throughput. I have tested Nortels AP and it uses the Atheros chipset which is a very goodPimplementation. If you also use Nortel's PCMCIA cards or any card with the Atheros chipset you could also run in turbo-mode which bonds channels to support up ti 108Mbs. Howesver note that this is proprietary and other 802.11a cards will not work on this AP. By using (3) AP's instead of (5) you can also use the standard channels (1,6,11) for the 'b' side of the system.

BTW, Nortel supports the Symbol Netvision phone and recently annouced support for Spectralink wireless phones as well. Nortel also has a softphone (i2050) that runs on a PC that is being ported to PocketPC clients. It will be out this fall.

hope this helps...