Forum

Hey all,

First post and it's a bit involved. I have a question about signal propagation in relation to power levels and would appreciate some insight.

This is being simulated in Airmagnet with 2.15dB Antenna and concrete walls with a -70 mask. I realize this may not be exactly true to life but maybe the images will help stimulate some discussion.

Lets say I have an AP at 100mW that has the following propagation.

[url=http://i.imgur.com/mmjEQ.jpg]http://i.imgur.com/mmjEQ.jpg[/url]

The same AP at 12 mW

[url=http://i.imgur.com/7KHET.jpg]http://i.imgur.com/7KHET.jpg[/url]

Is there any way you all reduce the "bad" signal from -70 to -100? Even with the AP powered down, the "good" signal is slightly reduced, but the signal from -70 to -100 still leaks to almost the entire building in practice and in this virtual simulation. I am trying to get a handle on this to avoid channel interference in 2.4.

How do you all prevent floor bleed? Do you just work with it? I get -50 through some floors but it is not reliable enough, and does not provide enough coverage horizontally to be of use. Is there a way to "flatten" the signal using an omni by increasing the power as described in the CWNA books? How do you plan that? Measuring on each floor?

[url=http://i.imgur.com/c3zHv.jpg]http://i.imgur.com/c3zHv.jpg[/url]

Thanks,
Cave

Hey Cave,
Thanks for the post, and sorry for our bad manners in not responding. You ask good questions that are at the heart of WLAN design, and they're at the heart of most WLAN problems. The unwanted signal that you speak of between -70 to -100 is impossible to avoid. Your illustration of decreased power is an excellent validation of the point that by shrinking cell sizes too much (microcell), there is a point of diminished return. In reality, the size of desired coverage is too small to be of much use, but you continue to blast interference throughout a building. Anyway, the solution to your problem is exactly what you're doing. Survey and plan. You'll never remove co- and adjacent-channel interference, but if you can optimize your deployment (with proper channel reuse, transmit power settings, and antennas) to minimize it, that is where you'll see the best performance.

This brings up a good point that we don't talk about enough. That is, since this "noise" is impossible to avoid, one of the best solutions is to use APs and clients with high quality radios and antennas. As radio quality goes up, a device's ability to distinguish between wanted signal and unwanted noise goes up as well. Of course, this is hard to measure. :(

One of the best ways to control your RF between floors is to use directional or higher gain omni antennas. You'll never prevent signals from bleeding between floors, but if you can find antennas where the propagation pattern is more horizontal and less vertical, it may be worth exploring. Look at azimuth and elevation charts to check propagation patterns, and if you can, compare these charts to the charts of your currently deployed antennas. One of the issues you'll find with this is that APs/antennas are mounted on ceilings, which means they're close to the floor above.

Marcus

Marcus,

You say that "As radio quality goes up, a device's ability to distinguish between wanted signal and unwanted noise goes up as well. [u]Of course, this is hard to measure.[/u]"

But isn't this almost exactly what you get with a radios sensitivity measurement ?

I can actually measure sensitivty in our lab and not have to rely on a mfg's claims. In the testing we do, the big differentiator has always been sensitivity i.e. better sensitvity = better range. I usually find that output power is very poor for predicting range.

At least for indoor and say 400' outdoor ranges.

Marcus,

Thanks for the well articulated response. When showing a customer a freshly designed and installed network, it is hard to relate the fact that there is going to be some channel interference in a dense mixed deployment.

I am trying to find the right balance between TX power and channel interference. If it is going to be there, and I can not avoid it, I would rather up my TX power to hit my db goals right? How do I know when I have hit that balance between tx power and channel interference? Are those the two variables I should have in mind? I see you mentioned TX sensitivity. I will look into that as well.

Thanks,
Cave

While AirMagnet is a wonderful tool, IMHO the algorithm that the s/w uses to determine co-channel interference is a bit out-dated. If you have a ?properly? designed WLAN no matter what the design criteria there will always be co-channel interference, except when coverage (RSSI) vs. bandwidth/capacity is the only design criteria. RF retries is a more important stat then co-channel interference, but even there the algorithm is out of date. I?ve seen AirMagnet say there was -72db RSSI/54Mb of WLAN coverage when a STA couldn?t get an ICMP ping reply. With modern radios co-channel interference is not the big problem that the literature suggests. Compare the results of AirMagnet to Ekahau in terms of co-channel interference, I rest my case. As to the planning functions of both s/w?s, neither allows for accurate RF reflection, so all simulation results are suspect. These are all s/w tools and need professional interpretation to be considered accurate. Please understand that I?m a practical WLAN engineer, I don?t care what the books say, I just know what works!!!

Hi Cave,
Sorry for the delay in response. As you might have seen on some of my other posts, I seem to be battling the same questions and design issues.

If you read my other thread you will see mention by Jackman of matching the client's needs in power. If you simply blast at 100mW you need to make sure your clients can at least match that and function well while doing so. I don't see any reference to what client types you are supporting. Are they latency sensitive applications?

Also what are those walls made of that it destroys wireless signals like that? If the pic is -70, is that what you are looking for as far as separation? If not, this pic does little in the way of determining cell size and placement. I know for the deployment I am working on we are trying to hit -76 for channel separation and -70 at the cell edge with 10% overlap as I recall. To be honest, I'll need to look up the overlap margin again however.

In 2.4 you only have the 3 channels but they could almost lay on top of one another and function for certain clients. (yeah I know not recommended, but point stands) If you are getting signals that die like that through walls, placement can be tricky, but you should be able to create a plan that allows for non overlapping channels to overlap in the halls.

As for the floor to floor, I actually set up an AP and did an active survey below it to measure the bleed through. What power are you setting the AP at that you can get -50 below it? What are your floors made of that it makes it through like that? Does that come down an elevator shaft or something?

Remember that predictive analysis is exactly that. Predictive. We went back and litterally tested an entire floor's AP placement to ensure that our predictive maps would be accurate. They were, and so we continue with predictive placement as it is easier to make changes virtually.

I hope somewhere in this ramble was some useful information for you. A little more information would allow us to give you a more focused response here I think. Client requirements and specs were one of the primary things I focused on in my most recent design. Making it fit in the space only came after gathering all the info.