CWNP Wireless Certification & Wireless Training

Your Account |

Search:

Surveying Mess....uh, Mesh.

Add to:

Written by Devin Akin

Monday, 14 July 2008

Site surveying has finally standardized. The process is well-understood among industry professionals - though not always implemented properly. Just when we thought we had it whipped, we get a curve ball: mesh. Have you checked out systems from vendors like Ruckus, Motorola, Meru, Cisco, and Aruba that have the option to have mesh APs connecting back to root APs? This is a seriously nice feature, but have you thought about doing a manual survey for such an implementation? Let's explore that, shall we?

Let's call him Bill. Bill is cranking along, measuring RSSI values - perhaps doing active and/or passive surveys using leading-edge site survey tools from companies like AirMagnet and Ekahau. All is well. Then, the supervisor calls and says, "Hey Bill, don't forget that not only do the clients have to talk to the APs at a level good enough for voice, but the APs have to talk to each other - sometimes over multiple hops - good enough for voice as well. Also, the customer wants some redundancy in case one of the APs dies, so makes sure each AP can communicate with at least 2 other APs with voice-grade connectivity." Bill just looks at the ground, wondering if this is even possible. This "one-day" site survey could now take a week. Place, test, move, retest, move, retest, move again. Horrible.

How is Bill supposed to make sure the connectivity level between the APs is voice-grade? Certainly not with traditional survey tools. Now we have to bring a controller or some piece of management software into the mix. Once the APs are associated and configured, is the portal AP (wired) seeing the mesh APs (wireless)? How many mesh APs can be seen and at what RSSI values? At what connectivity rates are each of the mesh APs connected to each other and each portal AP? If a client transmits a voice packet to a mesh AP and it has to hop twice to get to the portal AP, is the latency too high? What if the data path changes on the fly due to unforseen changes in the RF environment? Holy smokes batman! How are we supposed to survey that?

I'm thinking that there will be two schools of thought on this topic:

1) Implement mesh APs in hard-to-wire areas, as coverage-gap-fillers (to save money on wiring), and for some minimal level of redundancy. This will mean that you do a traditional survey, then hose it up "somewhat" by throwing extra APs here and there as needed for whatever purpose suits your fancy.

2) Design the network from the ground up with a mesh edge in mind. This will be done without doing a manual survey up-front, and the manual survey will be done as a verification-of-coverage-and-capacity tool only. Cisco's stance on survey's is that they are not for design, but rather for verification of a design. I think this line of thinking will play nicely with this type of implementation.

I'd love to hear some feedback on what you guys think will happen to site surveying due to this new technology at the edge.

Comments (3) Add Comment

Subscribe to this comment's feed

...
written by Reggie, July 13, 2008

I think when surveying or planning for mesh disabling legacy rates 1 and 2 Mbps, should be a standard practice , especially if voice is in the picture. Using these lower basic rates, is not as viable as it was a couple of years ago. Especially with 802.11n in the picture.

Cisco's approach is spot on. Site Survey's are validations of a design. I think the best design approach is to do a conservative predictive analysis using a planner tool. If the customer has deep pockets or simply wants a more detailed, cost effective WLAN, then a visual walk through to help understand the construction and existing spectral challenges is a must in a standard or mesh architecture.

Something designers need to consider with mesh:

Power to a mesh is always going to be an issue , unless we invent POA (Power over Air). When surveying for mesh here are a few questions to consider :

How will you power the mesh APs? (electrical outlets are normally two feet off the ground). New electrical outlets are expensive to install. APs are normally placed above the ceiling grid to reduce theft. (physical security). The customer needs to understand if they want this type of architecture it can get pricey.

If you are planning on running UTP to power the mesh AP, is there midspan switch or a 100/1000 Mbps Ethernet full duplex connection? If yes, is the mesh use viable?

We have with held from using indoor mesh based on the physical location of the power outlets and aesthetics of placing an AP out in public view--security.

...
written by Senthilraj CWNE#15, July 17, 2008

Its is always good to plan for mesh installation earlier rather than adding APs later. I go with Approach Number 2 .

Im not going with aproach number 1 .

Down below are few point that I wanted to highlight

1) Never use a single radio for backhaul and client access . We should use dedicated radios for backhaul and client access. If we mix up both, we will run into alot of throughput issues and finally mesh won't work. Throughput will take a hit even if one client gets connected at lesser data rates.

2) The backhaul radios should get connected to each other at least at a minimum link speed of 36 Mbps in an indoor network. SNR and another things should also be taken into account.

3) Always use 802.11a/n for backhaul and 802.11g/n for client access. I do not recommend 802.11g/n as an option because in an enterprise deployment, channel bonding will cause alot of interference.

4) APs with slots available to add on 1 or 2 radios can make the investment future proof, so when management plans to have a mesh, you add a radio and dedicate it for backhaul. The mesh on the edge becomes easier this way.

5) RF planning: Most of the mesh vendors have their own site survey tools, which will come in handy. A general site survey tool may not be as efficient in the case of mesh.

6) Single radio mesh should be used only in a SOHO network and not in the enterprise. Dual radio mesh can be used on the edge and for a pervasive wireless mesh networks. Multi-radio mesh is the typically the best option.

Please pour in comments on this tooooo.

...
written by Kris, August 05, 2008

I think the predictive tools are the way to go. Cisco has started using the predictive method but a company called Wireless Valley, now owned by Motorola, was doing it years ago and it works great. The LAN/Enterprise planner tools work excellent for both 802.11 and cellular. Mesh Planner works excellent for outside large scale mesh networks. The hard part is getting customers to pay for the cost of an initial site survey and predictive analysis. Customers do not realize how much back end cost they will save by investing a little extra up front. The savings can be tens of thousands or in large enterprises millions of dollars.

Initial pre-site analysis and prediction generally costs less than a full site survey does later down the road.

The tools mentioned above have built in site survey and optimizations tools that are excellent and ground breaking. I personally been using it to design and develop wireless installations for almost 4 years now with great success. (Assuming the actual install is done correctly by the contractors)

Write comment