Forum

hi all in every book it is sad that for affective roaming between adjacent cells, these cells must overlap about 10-20% but how this is done? how to calculate this overlap? this seems impossible

Modern site survey software packages can do that for you automatically. There are a number of vendors out there with high quality packages, such as Airmagnet. You input a floorplan of the area you wish to cover in one of several common "drawing" formats [ or you can make a drawing yourself in some packages ]. You can enter all sorts of information such as the type of materials used in the doors, walls etc, along with other parameters. A graphical output is given.

In fact, if you look to the right of this posting, you may still see an advertisment for a demo planning tool by Aerohive .... try it !!

It is possible to do it manually,by a very painstaking method, but it would take you forever.

Dave

Just as an add-on:

The issues of overlap etc are well covered in the CWNA book, and are well worth reading in the Site Survey sections. Not enough time to go into all the details, but you can use special software to get an idea [ usually quite accurate, provided that the input data is accurate ] to predict coverage areas etc for a building without physically having to walk through it. This will act as a guideline. however, it is strongly recommended that a physical "walk-through" is performed to confirm and "fine-tune" the predictive results. I have seen one or two places where a predictive survey was done, and the person responsible made some mistake s in inputting the data, then went and installed everything without a walk-through. That cost a lot of money.

With Airmagnet, you can "play about" with settings and locations of access points etc to get the results you want. I used the term "automatically" a bit loosely before, but apart from some mouse clicks and selections, the whole process is nicely automated.

Dave

Bye the way, this is a pretty good book about site surveying. Lots of good tips in it.

http://www.amazon.com/802-11-Wireless-Network-Surveying-Installation/dp/1587051648/ref=sr_1_1?ie=UTF8&s=books&qid=1264376165&sr=1-1

This one is a bit dated, but contains lot of good stuff on site surveys and an excellent section on wireless bridges [ generally poorly covered in many Wi-Fi books ].

http://www.amazon.com/Building-Cisco-Wireless-LAN-Syngress/dp/192899458X/ref=sr_1_4?ie=UTF8&s=books&qid=1264376267&sr=1-4


Dave

I wrote an entire white paper on this subject!

It's called the Fallacy of Channel Overlap. Here's a link to the article.

Basically - don't measure 'overlap' from an APs point of view, or from a 'graphical' nature. But design your wireless LANs to meet the client adapter's design specs.

You need to deliver a 'primary' AP's coverage, then a 'secondary' or 'backup' coverage from another AP.

This is measurable and can be verified, unlike the 'graphical' coverage ideas of the past.

Keith
http://wlaniconoclast.blogspot.com/2009/04/fallacy-of-channel-overlap.html

thank you all especially Keith. these notes that you need overlap lead some mistake, your first thought is about geometrical overlap not about signal strengths

longiroza

Yes, you're exactly right. I should have put that in my reply [ was going out the door and should have put more detail in ]. Usually I write volumes.

Take for example two AP's that overlap geometrically by 15 %. Unless you have an indication of the signal strength, the value of physical overlap is meaningless. One AP may be putting out a "reasonable signal" in the overlap area, whereas the second AP for various reasons might only have the tiniest of signals.

Imagine mentally [ don't have a graphic, but I'm sure someone does ] an x-axis and a y-axis. Y-axis measures signal strength. X-axis measures distance. Imagine one curve starting high at the origin [ zero distance ]. That curve [ sig strength of AP 1 ] would decay exponentially towards the right. At some distance further along, we would start to pick up the signal from the second AP [ AP 2 ]. This would increase from zero signal strength exponentially to some maximum [ right at the AP value ]. The whole thing would like a wide letter "U". There would be an area of intersection, and that is the key element.

And of course, as Keith says, it all "client centric".

Dave

Of course "The proof of the pudding is in the eating". It is vitally important to make test calls and/or walk around with a laptop or PDA or whatever and make sure that there is seamless transition from one area to another. The whole issue of Layer 3 seamlessness is another arena entirely, but having good solid Layer1/2 smooth transitioning is very important.

Dave

Imagine two singers at either end of a room both singing the same song at the same volume at a party with background noise of people yapping.

A old man and a young guy want to listen to the music without interuption as they move across the room from one singer to another.

Each will gave a transition area where they can move from listening to one singer to the other.

Because each has a different sensitivity in hearing [ young kid better than old man in this case ???¡é?¡é???????|.sort of analagous to data transfer and some other thing like VOIP ] their transition areas will be different. The young kid will able to tolerate a bigger degradation in signal level before ???¡é?¡é?????¡­?¡°moving???¡é?¡é????????to the next singer [ AP ]. The overlap areas will depend on signal quality for the application [ amongst other things ].

Knowing application ???¡é?¡é?????¡­?¡°tolerance parameters???¡é?¡é???????? becomes very important.

There was a lot of psychology involved in the development of digitized voice and video [ how the brain interprets voice , where did an 8khz sampling rate come from, how does CELP work, how does the brain interpret video images, how much bandwidth do we need to fully represent voice and why ? ???¡é?¡é???????|all highly important stuff for VOIP . ]. If I get time later, I hope to put up an audiocast on some of this stuff.

Dave