03-19-2008 08:19 AM - edited 07-03-2021 03:33 PM
I'm curious:
a) do most people here have site survey kits and if so b) what's in them?
Which AP do you use? What antennas do you consider good enough to keep in the kit? Do you just use a lot of extension cords, or do you have some sort of battery system rigged up?
03-19-2008 09:59 AM
My site Survey kit -
It consists of a cart - maby 2' by 3' and 4' tall.
We have a painters pole that can extend to 15'. The painters pole is mounted to the side of the cart. we have an AP mounted on a flat piece of wood, which in turn is mounted to the end of the pole. This way we can raise / lower the AP to the celing for testing. On the bottom of the cart is a UPS. We ues a rack mount variety, so it just sits on the bottom of the cart.
Equiptment - LAPTOPS, LAPTOPS, LAPTOPS. can never have too many laptops or laptop batteries. We use 3 different laptops - 2 with built in wireless, one we stick PCMCIA cards into. All have their power bricks hooked up to the UPS so we can charge batteries from location to location.
Digital camera and accurate maps of the location.
Software - Netstumbler, Kismet, EXCEL are all great tools. Try to print a grid on your map and take readings at specific points, then draw signal patters on the grid.
Most of the installs I do are with omni antenna, I try not to switch the antenna up to much, it just gets to be too many variables
03-20-2008 05:19 PM
Proper site surveys are the result of auditing your network design by simulating the network to as close as possible to the one you intend to deploy in production. I suggest that you have a good site survey software such as Airmagnet Survey Pro or Ekahau. You then need a variety of antennae including omni directional as well as directional patch. Use the antennae that you intend to use in deployment. Remember, you should never exceed 20% of your APs with omnis as this will drastically increase the noise floor and create an unstable RF environment. You need at least one of the APs that you are going to deploy and a battery backup to power it. 2 of each of these is preferrable to show proper cell overlap. A cart to roll your client device on is the next thing needed but not required. Make sure you survey for the worst client you intend to deploy. If that is an RF gun then you should use the RF gun to verify distance and reliability. If it is a laptop then use a laptop. Use the output from the survey software to validate designed access point locations. Good luck.
03-20-2008 07:50 PM
"Remember, you should never exceed 20% of your APs with omnis as this will drastically increase the noise floor and create an unstable RF environment."
Dennis, this is an interesting comment. Can you elaborate? Are you saying that over-omnidirectional coverage patterns create too much overlap and co-channel interference? Don't indoor reflections offset much of the gains (pardon the pun) of directional antennas in all but open-floor environments?
Thanks in advance for your reply. I'll award points once I hear what you have to say.
03-20-2008 09:10 PM
"Remember, you should never exceed 20% of your APs with omnis as this will drastically increase the noise floor and create an unstable RF environment."
Would the Auto RF... be any help for this unstable RF envi?
Another 5 points here waiting to be submitted...:)
03-21-2008 05:01 AM
When transmitting an RF signal we must always remember that the ussable signal transmitted is only a small amount of the actual transmitted signal. The unusable part of the signal travels much further. Imagine it as an archers target with the bullseye being the AP and each concentric ring going outward being a reduction in usable signal. About half the distance or more is unusable but does effect adjacent areas as noise. Omni directional antennae spread this noise over much larger and uncontrolled areas. The majority of noise floor issues in complex RF deployments are a result of native APs with omnidirectional antennae. By starting your deployment design at the perimeter of the coverage area and using low power with directional antennae at a slight downtilt (amount based on the actual environment), you control the amount of electrons emitted and the pattern and area they cover. Fill in the central areas of coverage with omnis also powered low to complete the coverage area and odds are you have reduced your noise floor by 15-20 dBm. Auto RF will lower the power where necessary but can not compensate for the wider areas of unusable noise propagated from an omni directional antenna. Make sure when using direction patch antennae that you use either a diversity patch or 2 non diversity patch antennae to assist with mitigating multipath distortion. Above all, only use the power required to cover the area. Low power is your friend. High power can be your worst enemy. One other thing to consider is the environment that is mentioned in a post above. In very tightly walled areas such as offices and hospitals you need to account for areas where aesthetics and functionality require the use of omni directional antennae. In most cases these are multifloor environments and if omnis must be used the power is really critical here as well as staggering like floors so that no two floors antennae are directly one above the other as this will cause noise and cochannel interference between floors. Again, I like to use directionals where possible in this enviroment as the downtilt feature of 10-15 degrees will usually result in the RF signal remaining on a single floor. Diversity in this environment is critical and I really suggest using 2 non diversity patch antennae at about a foot apart to mitigate multipath.
Good Luck!!
03-21-2008 06:05 AM
Thanks Dennis,
I agree with low transmit power and would add limiting data rates as well. This 20% rule -- can someone produce a document explaining it? This is a pretty controversial concept. The goal of RF containment is implicit in the rule. But what about things like effective client handoffs/roaming, which benefits from coverage overlap and smaller cells? What about coverage hole mitigation and coverage resiliency? This sounds like it may be the new paradigm to support location (coverage from the outside in).
Anyway -- 5 points to whomever can produce this rule in writing somewhere.
03-21-2008 05:36 AM
I agree with Dennis and I also disagree. We have a large hospital with just over 400 access points installed and working. They are the Aironet 1131 series access points fronted by 3 WiSM controllers. We deployed the 1131's for a couple of reasons, they look good, they mount below the ceiling for easy access, and we can see the lights on the unit to see if it is working correctly. Also, they are cheap!
I have gone to several site survey classes, been involved in some very complex site surveys with Cisco and I understand the concept that Dennis mentioned. But I think that there can be a successful deployment with just one type of antenna.
I attended a Cisco VT conference about a year ago in San Jose and one of their top survey guys spoke about the 20% rule. After he spoke, I had a conversation with him and explained how we had used only the 1131 with the omni and had experienced great success with it. He went on to say this... "that's great! We have a guy that believe's in the omni's so much that he will only deploy omni-directional antennae in his surveys. I think there are alot of right ways to do it."
With that said, I think there are also alot of wrong ways to do it as well. Getting a survey from an experienced engineer is the way to go, even if they don't deploy it. And I can say we have used Dennis recently to make sure what we were doing worked. He has looked at our wireless with great detail and made recommendations that we will use moving forward.
03-21-2008 06:19 AM
Beav,
Thanks for your reply. Can either you or Dennis recommend site survey classes or publications? I am intrigued by the directional antenna emphasis, and the noise floor issue. Might make for some more Xirrus sales ;)
03-21-2008 06:32 AM
Bruce,
Sure. Global Knowledge always offers highend Cisco classes in advanced wireless. Here is a link to the wireless section of their site.
http://www.globalknowledge.com/training/category.asp?pageid=9&catid=206&country=United+States
I just took a look at your site (www.partners.org) and am interested in your role there. I am always curious if others in healthcare are experiencing the same headaches as we do. If you get a chance, drop me an email at dabeaver@srhs.com. It would be great to talk IT healthcare with someone in the same arena.
03-21-2008 06:38 AM
Oh yeah, we use GK a lot. In fact I know one of our guys took one of those classes and they promoted use of AP models with embedded omnis. I don't think their trainers will promote the 20% rule.
03-23-2008 08:40 AM
I will try to get Cisco to post Neil Reid's document on the 80/20 rule for Omnis. I do agree that are some cases where Omnis are the better fit. However, you can easily attain proper cell overlap and coverages utilizing the 80/20 rule. Like I said earlier, there are cases where the aesthetics police won't allow the use of unattractive antennae. The old Airespace APs (Cisco 1000 series) served this market quite well. The 1020 or 1030 being the AP of choice for multifloored deployments. I know, I've heard all the arguments against them, but 16 SSIDs out of the box. Bi-drectional antennae with the proper 15 degree down tilt, and exterior antenna connectors if needed far outweighed their negatives which were basically low memory, no console port (A FIPS requirement), and some very freaky spatial diversity. If I can't get Cisco to post Neils docs then look for the coming book that I am currently working to complete about proper antenna propagation, proper use of power, the use of gain, and antenna placement.
03-25-2008 07:40 AM
Dennis:
Good posts, great information, one request please:
Put some white space in now and then. Block text in the little teeny tiny Cisco font makes it really hard to read ('specially for us old folks).
Think of it as "taking a breath" as if you're speaking.
Please don't take it as criticism, that's not the intent, I only mean to improve readability of some excellent information.
Thanks!
Scott
03-25-2008 03:45 PM
Thanks Dennis,
I look forward to this doc and your upcoming book. Controlling downtilt and the impact of elevation beamwidth are nuances to the survey process that are not common knowledge.
Regards,