Blogs

Two ways of designing Wireless LANs.

By keith.r.parsons posted Sep 04, 2014 03:17 PM

  

The first, and perhaps the easiest, is to look at designing your Wireless network like a big calculation. This begins to get quite complex with many variables, yet the idea behind it is simple enough.

 

Build your design by looking at a variety of factors that go into how many access points you might need. If you look at enough variables, it starts to feel like you might have a good handle on getting the proper result.

 

Let’s go through a simple example.

 

First, you have to pick the Access Point of choice. This will determine the capabilities and capacities that will be used further on in the calculations. Things like single band vs dual-band, supported channel widths in 5GHz, number of supported associations per radio, how many concurrent SSIDs, what you Basic Data Rates will be. This will give you available airtime to support client needs.

 

Next, you take into consideration detailed information about your client base. How many, what type, what 5GHz channels might be supported, channel width, supported spatial streams, supported MCS rates, etc.

 

Finally, you add into your monster calculation the types of applications and services that will need to be supported by your clients. TCP data rates needed, RTLS, VoIP, video conferencing, streaming video, etc.

 

Put all of this in a big massive spreadsheet – along with a bunch of assumptions and out drops a number of Access Points you’ll need in order to support those assumptions.

 

Don’t get me wrong, this is a great way to get a quick estimate and get in the ballpark with a Bill of Materials and a starting point for Access Points.

 

But it is still JUST A CALCULATION – it has nothing at all to do with the real world. Especially with the RF situation you are designing for.

 

The REAL way to design for Wi-Fi that just works, is to design around RF Contention areas. The total capacity of any high-density WLAN is totally dependent on being able to get the absolute most Frequency Reuse as possible.

 

Again, to reiterate: It is all about Frequency Reuse!

 

No matter what your calculation process yields in the way of Access Point counts… it all comes to naught if you can’t manage to get frequency reuse.

 

Now what do I mean by ‘Frequency Reuse’. In any given area, how many times can I reuse a specific frequency? Think of a large circular amphitheater, perhaps 100m across. (imagine about the size of an American Football field only in a circle). With all the seats arranged in concentric circles around a tennis court in the center.

 

How many Access Points will you need for Coverage in this arena? Think hard for a minute…

 

OK, you’ve had enough time. The answer is ONE!

 

You’d only need a single access point in such a facility and with an Omni antenna and high power setting (still below FCC limits) you could have all seats covered with say -70dBm of signal. Meaning, any seat in the arena could have someone sitting with a smartphone and have Wi-Fi access from a single Access Point.

 

That’s the way omni antennas work. And with nothing in the way between the Access Point on center court, and each and every seat. You’d have fantastic coverage. No seat would be more than 60m-70m from the AP and life would be great.

 

Well… not totally great… because you would be limited by the association count capability of the access point. Perhaps 250 total clients spread across two radios and you’d be at association max. This is nowhere near the amount of seats in the arena.

 

So the answer for Coverage is ONE… but that doesn’t meet the capacity requirements at all.

 

So, you triple the count and put three Access Points on center court. You’ve now got a radio on channels 1, 6, and 11 and you’ve tripled, the capacity on 2.4GHz and tripled the capacity in 5GHz with radios on channels 36, 44, 52.

 

But you are still nowhere near the capacity goals for number of access points that came from your monster calculation project.

 

Thus we run into the REAL design issue. How to get the most number of Access Points without causing co-channel interference. Because as soon as you have two AP’s on the same channel, they both see each other, and the clients see each other… because of how 802.11 works – they will all defer to each other in a large contention domain. Thus you added an extra access point, but didn’t gain any additional capacity – all the access points and clients on the same channel will share the bandwidth together.

 

This is why I try to emphasize it is all about Frequency Reuse. The goal of Wi-Fi design is to get the most number of AP’s you can without causing Co-Channel Interference and contention areas.

 

Thus, instead of using only calculation tools, you need to have in your toolkit arsenal, at least one tool that can do predictive RF analysis. You can use these ‘survey’- type tools to predict where your RF is going, and how to move AP’s around, choose different antennas, different power settings, and learn to use walls, floors, and other RF dense obstacles to design IN RF frequency reuse!

 

The next time you are thinking about Wireless LAN design remember… It’s not about the calculation as much as it is how to design IN Frequency Reuse.

7 comments
3 views

Comments

paulrfinlay
Sep 12, 2014 10:30 AM

I've used WCS/Prime, Ringmaster, AirMagnet, Ekahau and AeroHive Planner and I've found they all provide a remarkable similar prediction of RF coverage. Generally I find that the most important thing, as you say, is to ensure that the assumption used in the planning tool are accurate - which includes building material attentuation and AP transmit power and gain. Garbage In Garbage Out

revolutionwifi
Sep 05, 2014 09:37 AM

Joeri, I love your comment! I agree that some on-site sampling can really help to verify the accuracy of the RF predictive modeling. As Keith said, you spend a bit more time and it costs a bit more, but in the end I think you end up with a better installation that requires less "fixes" or "tuning" after the verification site survey is performed. The trick is to figure out just how much on-site sampling to do, which comes with experience in my opinion. Personally, I look try to perform a few samples in every "type" of coverage area in the design, since often times there are multiple types of areas (think classrooms, hallways, lecture halls, gymnasiums, office cubicles, stairwells, outdoor areas, etc.). I definitely don't do a full AP-on-a-stick pre-installation site survey, but I do try to ensure that all of my main assumptions in the predictive modeling tool are correct.

 

On a related topic, what predictive RF modeling tool is everyone's favorite, that they feel is the most accurate? There are many to choose from, and I'm just wondering if people have tried multiple and found any specific one to be their favorite in terms of accuracy (obviously we'll also have favorites based on familiarity and training, but I'm more interested to see if anyone finds one more accurate than others).

 

Cheers,

Andrew

keith.r.parsons
Sep 05, 2014 06:42 AM

Thanks for your comments. I like your style!

Not only a post install verification survey prior to tuning before turning
over WLAN to customer? that alone is something I whole-heartedly agree with.
But also your take on validating the predictive design while on-site. A bit
expensive, time-wise, but it will make for better designed WLANs.

Thanks again for your comments.

Keith

joeridewinter
Sep 05, 2014 03:02 AM

Great post, from my experience I would like to add onsite verification tests of the predictive RF analysis. It’s always recommended to do a verification site survey after the install and before going live to fine-tune the RF, but I like to do some tests onsite with the chosen directional or omni antennas before install to verify my predictive analyses. Predictive RF analyses tools do not take in account (at least from my opinion) RF reflections certainly for indoor high-density deployments and even for outdoor deployments. I am sometimes surprised how far those side and back lobes go, don’t underestimate the side and back lobes of directional antennas in high density deployments they can really mess up you’re frequency reuse plan in contradiction to you’re predictive RF analyses. Even how good you think you’re predictive RF analyses is in the real world the results can be different, to avoid to come to this conclusion after the installation I would recommend to do some onsite tests with the chosen antennas and calibrate those results into the predictive RF analyses. Of course the more you do this exercise the more experience, probably you know the ins & outs of you're predictive software and know how to tweak it to get the best results, at the end it comes all down to knowledge & experience ;-).

 

Cheers,

Joeri De Winter

KeithRParsons
Sep 04, 2014 11:45 PM

When writing this post I was thinking back to many expereinces I've had with folks who used ONLY the first calculation method to come up with number of AP's then went directly to the installation phase. Then later when the Wi-Fi was having issues, upon technical analysis, we found there were way too many AP's for the given area. The solution was to turn off AP's to improve the Wi-Fi performance.

 

Rinse, repeat - move to another site and do it again... 

 

I've spent years earning a living fixing WLANs where they didn't look at the RF side of the WLAN design process. So this 'attitude' must of slipped out in my post.

 

Just to be clear... I think the best WLAN designs come from using BOTH techniques. Both are relevant, and both are necessary. 

 

So I totally agree with Andrew's quote: "Bottom line: WLAN design involves BOTH capacity planning AND coverage planning! Not only one. Not only the other. But both."

 

Thanks for the coment Andrew.

 

Keith

Chappy4756
Sep 04, 2014 06:44 PM

I would think that capacity planning provides a benchmark for measuring performance requirements and whether the design is successful. Frequency re-use, I would definitely agree is probably the most important factor and consideration in how you can deliver that capacity, performance and quality client experience. Especially in high-density installs. My thought is that both are mandatory parts of a proper design.

revolutionwifi
Sep 04, 2014 06:23 PM

Keith,

I think you over simplify the situation in your example. We need BOTH capacity planning AND frequency re-use through coverage planning! They are not mutually exclusive, doing only one or the other are both recipes for failure.

 

Performing only capacity planning calculations, as you have stated, is a predictive exercise and cannot account for co-channel interference issue of coverage planning.

 

Performing only coverage planning, which you have not discussed, give you coverage everywhere you need it while avoiding as best as possible co-channel inteference. But it does not provide any accurate method to tell you how many APs on non-overlapping channels or coverage areas you need in the first place. Do we need 1, 3, 6, 10, 15 in your example? We could design in 20 APs with directional antennas and great frequency re-use and minimal co-channel interference, but still have no clue if capacity needs are met despite having great frequency re-use.

 

At best most RF planning tools try to guess at this with square footage rules-of-thumb such as 2,000 square feet per AP for high capacity. But this is just a guess and there are many high capacity situations where APs need to be colocated and coverage rules-of-thumb break down and don't provide sufficient capacity. There are also low capacity situations where these rules of thumb can create a design with too many APs if used improperly. Without capacity planning network designers don't really know how many APs they need to meet capacity and throghput demands.

 

Bottom line: WLAN design involves BOTH capacity planning AND coverage planning! Not only one. Not only the other. But both.

 

Frequency re-use is a major factor in WLAN design, probably THE most important. But it's not ALL about frequency re-use. That overly simplifies a complex task. I would say "It's all about frequency re-use to deliver the correct amount of capacity!" 

 

Cheers,

Andrew von Nagy