Forum

Having installed and managed one of the world's largest campus WiFi networks starting in 2001, deploying 200 buildings, I can tell you what we found:

1. There are some businesses selling surveys to make money. We had 200 buildings to do and for one buildings, the firm we initially partnered with wanted $25K for the survey alone in that building! We showed them the door and used the money to setup a project team to do the installations from that point.

2. Our installation team spent time doing surveys in the first few buildings and then just did spot checks on buildings characteristics and went from there. As many found, it was easier to spend the money on more APs that time consuming surveys. Having said this, we cover surveys in detail in my classes because you can't eyeball it without experience and surveys is a great way to get experience.

3. We had dozens of new buildings being constructed (at the building plan stage), a little hard to do surveys. Users expected wireless the day they moved in. We later did predictive surveys on some of these sites and found that an experienced person could do them manually faster by looking at the building design. Most software (some from some large companies) don't even important autocad files or even look for lines on images; some poor person has to redraw all the walls, windows etc. Very time consuming to get pretty color pictures. Impressive for the a report but deep down, it's not truly predictive or efficient unless it can understand and autocad file and make intelligent decision based what it sees.

4. We found post-installation surveys *critical* to confirm coverage and adjust from there. Ensuring proper AP density is key, dealing with manual frequency and power is a waste of time except for that moment in time.
Stuff is very forgiving except for high usage and VOIP.

Systems of any sizes have to be automated. What seems to be missing is a tool to do the "walk about" to confirm coverage and communicating back to "mother ship" for auto-configurations. Walk abouts (clicking on the map where you are) will likely always be important but the rest is automated except for small systems in the middle of no-where.

5. In business districts and with the crop of new muni-wireless and city-wide wireless networks going up, the precious survey done to allocate the channels and frequencies in a location is out the door. What really needs to happen is 802.11h taking off for all spectrum to ensure that devices only spew out the required power to reach APs and that AP densities increase (IMHO). For muni installation, GPS roll-about will be important to deal with access and mesh performance, at least for outdoor verification and indoor predictions.

Having been responsible for 1700+ APs on a network, you can't deploy or manage large networks without automation. Municipal wireless networks will confirm that in my opinion. As you start having APs on street lamps, automated systems following mandatory regulations will be key in having a usable, cost effective system.

I really hope that the regulatory domains continue to mandate TPC and DFS for new spectrum; that's the unlicensed's spectrum knight in shining armor! Although some don't agree, I think the move to implement this for 5GHz is fantastic!, would be great to see this in some lower frequency unlicensed band in time for 802.11n MIMO... That would allow a higher EIRP as well...

With that, we have 4G wireless and don't have to 5 years to pay outragious prices for it. I won't even comment on EVDO or HSDPA (aka 3G ) today... :-)

Just some thoughts...

...................................
Jonn Martell, PMP, CWNT
Wireless Consultant, www.martell.ca