There are lots of discussions in the wireless industry about design. Should we do “AP on a stick” surveys for every site? Are passive surveys useful? Is there a place at all for predictive designs, or is there some combination of all of these that we should use?
Should We Really Be Doing Wireless Designs at All?
There has been a lot of press on some networking sites and all over Twitter and other outlets recently that wireless surveys and designs are no longer needed. There is a perception that controllers are now smart enough and that access points can just be installed, and the controller will do the rest. Then if any issues arise, just throw up some more APs and all is well.
This perception could not be farther from the truth.
Yes, controllers have come a long way in the last 20 years or so, and yes, they can make up for a lot of short comings in designs, but should we really let them? This reasoning is flawed in so many ways. Controllers cannot understand what the environment looks like from a standpoint of walls, attenuation within the environment, as well as user density and capacity needs. Controllers simply look at the environment from what is being reported back by APs and clients. So, adding more APs and introducing more spectrum usage and interference is counter-productive when there is no baseline to even know how the environment looked to start.
Predictive Designs Can Save Lives
Well kind of…
Predictive designs of a site can seem tedious since you really can’t see what a site looks like until you get there and actually start getting readings. Despite this, a predictive design will at least get you close to 85 percent through the design.
When a predictive survey is done and implemented, validated reference guides based on system requirements can be used to get a baseline of where the wireless design needs to begin. Here is a good spot to start with guides from Aruba.
Once the requirements are understood by the customer, using survey software, the floor layouts and building set-up can be imported. This is crucial to see cross-floor bleed/interference. Walls can be drawn, and we can start to see what the RF environment might look like as the APs get placed.
When the APs are placed in a predictive model of interference, based on channel plans, power settings and the like begin to show in the software. Then, depending on the software being used, capacity requirements can be input, and devices, applications bandwidth, and so on can be set up. Now, we have a clearer picture and can begin to develop where additional APs are needed to handle client capacity – especially for bandwidth-intensive applications. But we need to keep in mind this is only a prediction. Real-world measurements still need to be made.
Once this predictive modeling is complete, the design can be taken on-site and validated by an engineer by taking attenuation readings using an AP on a stick as well as possibly a survey of the current wireless network. This will help determine any weaknesses or interference to be cognizant of as the new design is deployed. Adjustments can then be made to the design as needed for optimal design and deployment.
Finally…
With wireless design, there is no end-all, be-all. There is always a combination of tools, skills and knowledge that should go into a well-designed and deployed network. A predictive design is just one of these pieces, but it can be a piece that could potentially save your bacon when you get onsite to deploy a network.
Related Content
The Aruba Architecture
Aruba Validated Reference Design Guides
Read My Other Blogs
Mama Says Channel Bonding is the Devil
