I would love to hear from anyone who would be willing to share their Wireless survey methods. Any thing from Active/Passive, AP on a stick, Outdoor surveys, challenging environments, new installations and existing ones.
Ive my own real world experience, read the books and done training but sometimes its great to exchange that info to get tips, alternative practices etc with peers :-)
Look forward to reading
I used to do the AP on a stick method with the CIsco AP on half power (11) however, It was always for a large installation in which a wireless controller would be using Radio Resource Management to determine power levels and channels.
Until you install all the AP's, you really don't know what power setting they will be at. You just have to plan / design the best you can.
I've spent most of the last decade specifically teaching this.
First - you 'may' want to do a pre-installation survey, but you MUST do a post install verification survey. That's the one that is most important. I can't tell you how many times I've been called in to fix a problem that was caused by the installer not testing after installation.
Second - going on-site isn't required. Highly recommended. But not required. Especially if you are doing an on-site post installation verification survey. You'll be able to solve issues at that time. Since you have to be there anyway...
Doing a full 'AP-on-a-Stick' is so old fashioned. Sure - you need to know about the building's materials and potential RF dead zones. But designing is different than surveying.
Most important DESIGN to meet ALL your system goals. Coverage is only one factor in a good Wi-Fi design. An important one - but only one of many you should be designing for. By the way - it's also the easiest to design for. As I say in my WLAN Design classes - Coverage is Easy! (at least 20 times in three days...)
The real killers of your Wi-Fi design, and the most difficult to design around, are Co-Channel Interference and Capacity.
I've put together some white papers to help with these - feel free to download them here:
Seven Rules for Accurate Site Surveys
The Fallacy of Channel Overlap
False God of dB
How to Cheat On A Survey
Designing for Voice
Learning Wireless LAN Technologies
Quick & Dirty 802.11n Design
Quick and Dirty 802.11n Design-2
Using The Right Tool for the Job
Want, Don?t Want,Don?t Care: meeting design specs
WLAN Counter Intuitive
If you have any questions - drop me an email
[quote]As I say in my WLAN Design classes - Coverage is Easy! (at least 20 times in three days...)[/quote]
Yep, I've never even met Keith (yet) and I've seen him write that and say it, back when he had a podcast, at least 15 times. :) It's an important point so it definitely warrants repeating.
Personally, I tend to agree that a post-implementation survey is very valuable. However, for those instances when I need to do something before deployment I like to do two things:
1 - A passive survey so I can see what's already 'in the air'. It's good to know what kind of wifi and non-wifi interference you'll be dealing with.
2 - A predictive modelling of the environment using a tool like Ekahau which takes building materials into consideration. More importantly, it takes performance and capacity into consideration as well. Will it be perfect? No. But it will give you a good idea of what you should be doing and is certainly better than eyeballing it.
Your post-implementation survey will either validate the predictive model or point out areas that you need to make modifications/adjustments (both to coverage and performance/capacity).
Keith is a legend in this field, and his podcasts and notes are very helpful.
The main issue I come accross is generally poor project management methods before you get anywhere near looking at a survey. If the company asking you in don't have a clearly defined project structure, you are going to inherit problems and liability.
If this is the case, you have to define the project framework for them, I recommend looking at CompTIA Project+ as a great primer. It's based on the PMBOK, as is Prince and PMP, the latter two both needing documented project time logged so are quite advanced, whereas the former (Project+) is entry level.
The longer a project proceeds, the more difficult it is to affect the outcome.
You need to build in the cost for doing the planning, pre survey and post survey and ensure it is done with a view to the overall specifications.
Once you have that right, then you can get on with expert surveying as per Keith's recommendations.
Like Keith said "designing for your system goals" is the most critical result of the survey. Keith is the best I have ever worked with period end of story, the step to meeting the system goals starts long before we get out a floor plan out though. Chapter 1 of the CWDP book talks about finding out your customers requirements. This can be one of the hardest processes, since most of the times the customers need to be educated and coaxed to give that information. It might take multiple people from the customers team, to deliver this information correctly. This is the part of being a WLAN Professional that some people miss. Once you have the requirements nailed down you can pursue a desktop planner design, a feild survey, but a post verification survey is mandatory!
I personally like doing a desktop design using AirMagnet planner, with validating some key variables before hand. I like to go onsite with an AP using survey in passive mode and gathering what my loss is through bearing walls, exterior walls, doors, firewalls, and looking for any gotcha's. The gotchas are usually where they have built an addition to the building and they leave the old exterior wall in place and butt the new exterior wall to it. This also gives you more face time with the customer though, to glean any type of information you can, any time spent with customers is a good thing. Once I have these components my design will be more accurate.
With AirMagnet Survey Pro version 8.2 and above, you can do multiple surveys at once. A passive is always mandatory to "see" what is going on in the air. I also like to do an active or a perf survey also. understanding what information is being delivered on the survey, is a whole other dimension. Using features like "Airwise" on survey pro to make sure that your system goals are met are extrmeley valuable to your success. Most of all time in the field and the knowledge from people like Keith R Parsons and the CWNP program are priceless in your success though.
Good luck on your surveys.
So many use cases and sometimes so little time for testing! I've recently been tasked with bringing about a Wireless LAN global infrastructure project to life and complete implementation of multiple designs/scenarios all within two years. Approximately 10 million square feet of space is up for design/redesign and around 2 million of that i have surveyed by myself. The bottom line is always YOUR customers bottom line, which as quote above sometimes is defined as asking an unknown [insert variable] amount of people for tidbits and maybe finding out which tidbit is the biggest tidbit. I ask about devices that will be allowed to connect and examples of their profile / application analysis data available for (typically) wired based access. Once i have that data and where that data is most likely to be seen (professional offices, call centers, retail locations, road tours, work at homers, executives,etc) i either go with a high density fits all (worst case) or a per use case design approach. If money is not a problem, pick what your customer states may be the best for them. This also speaks for IP subnet planning, dhcp leases/timers, policies for ingress classification/egress marking and the whole shabang since each of these details could end up defining or at a minimum contributing to your design goals. why does this matter for site surveying? It may matter, especially f you are looking for a mission critical, SLA driven solution that may scale, then the details matter. It's probably more challenging if you were like me for many years where you are passionate about 802.11 solutions but were not given nearly 100% of your time to focus on it. Back to the surveying spiel! I used to produce a predictive survey for a reference (made me feel better) with Airmagnet and using what i found in plenty of real world surveying to be a sound number to produce capacity, coverage and weakest link accomodation designs. Then, once onsite, i would take another walk and make sure it all looked right--did the sq footage and distance between APs "feel" like it "looked" right, did mounting of APs present any changes due to challenges with location? Once it all sort of felt "better than wrong" i would place my first AP on the ceiling. If i was using an autonomous AP, then i would just POE it up and setup my power and channel (post RF survey and quick channel scans with AM WLAN Analyzer) and if i was using a lightweight AP, i would do the same but just a location to put controller, switch, power and other tools for the survey needs. The pre-survey work for me entailed the following: iPerf/jPerf server (standard client laptops, drivers, TCP windowing - make it real); on the main PC/laptop acting as a server i would run remote terminal services to control the iPerf/jPerf client commands to my test machine "server", SSH daemon, sFTP server and ICMP enabled. I would take my test machine (usually one of the most used machines in their environment) and start an SSH session and ICMP pings to the server and then using the server machine start the iperf -c test to my machine as a server. I would at the same time run AM WLAN Analyzer on a seperate machine that i placed on a cart to observe the statistics of my associated client (using no link layer encryption either) and based on a quick run around the area performing these tests, i would make the decision on whether or not the AP was in a good location or not. Running reverse direction (client to AP direction) tests with iPerf were important obviously too. Once i saw the desired results, i felt good that this client both in upstream and downstream had no issues at a given cell area relative to this AP and the radio + channel being tested. (I gave up surveying for 2.4Ghz unless absolutely necessary--lucky to not be an issue where i am anymore i guess.) These tests took approximately 10 minutes and helped solidify that the AP was at a good starting point. I would then do a active or passive survey (client explained the differences and they made the call) with this AP (using the AP channel in settings) and walk to the edge of the cell or as far as i could stand (no less than -80dBm RSSI) if i had a huge amount of space to cover that night. At least with a passive you don't get as many drops with the survey software. I used to use the active survey all the time but with the thorough testing i do and now adding multiple devices to the test (prior to surveying) i can do passives. At some point i realized how this may be a waste, regarding doing all of the testing and channel observations (retries, data rates, rate shifting, signal distribution, throughput in tx/rx kbps,etc.) so i recommend doing all of your hardcore hammering and baseline testing (with bulk of your clients for sure) and then if you know that your AP locations are actually working, then you can do a final active or passive (i'd go with passive) survey. What do i get from this? I get a darn baseline..i guess i'm old fashioned. If i do this method, say using 3,900sq feet per AP, with between 1 - 5mW power output on a 2.2dBi gain dipole or internal antennas, then i know that at a mimimum Airwise tells me that i met my goals for coverage, capacity and a few others. It is important to see the up and downstream data but putting your time and energy into an active survey with static settings, to me, honestly gets tossed when your use automatic features of AP and Controller deployments. Hope i did not write some conflicting detail here as i tried not too!
On to User Acceptance Tests!
p.s. (Keith and i met 6+ years ago and i've had the pleasure of bugging him off and on ever since via email, phone and twitter- he's a heck of a mentor..even my wife knows who Keith Parsons is and she certainly is not an 802.11 enthusiast.