Subscribe to Newsletter
Tell a Friend
Print this Page
09/27/2007Are You Dense Enough? 40 is the New 20 in Metro Wi-Fi Nodes
 Node density of Metro Wi-Fi infrastructure is increasing.
The early metro-scale Wi-Fi hype was propelled by the notion that Wi-Fi was a low cost, alternative infrastructure. The cost of that infrastructure is driven by the number of nodes (access points) required in a given area and the number of backhaul connections required to interconnect that network. The rule of thumb a few years ago was 20 nodes per square mile for a suburban metro-scale Wi-Fi network. One vendor asserted it, and all the others had to agree that they could do it, too. It turns out that this density is way too optimistic for most cities.
Novarum has been testing broadband-wireless networks for more than a year for our Wireless Broadband Review . Our testing shows that higher node density is required for metro Wi-Fi networks. Higher node density leads to better service availability and performance. Better service availability leads to more subscribers.
The average node density for metro Wi-Fi networks tested in 2006 was 40 nodes per square mile. This higher average was influenced by Toronto OneZone, which is 105 nodes per square mile. In the urban core of Toronto, the OneZone network is very high performance and supports indoor coverage several stories up in buildings. While this seemed very exotic at the time, we now believe that that is the appropriate model for urban metro Wi-Fi networks. Even If we remove Toronto OneZone from the calculations, the average node density for metro Wi-Fi in 2006 is 30 nodes per square mile.
In 2007, we retested several networks including Google FI in Mountain View and EarthLink Feather in Philadelphia. The node density in Mountain View increased 13% to 43 nodes per square mile. Philadelphia increased 55% to 48 nodes per square mile — double their original estimate. The trend is higher node density for Metro Wi-Fi networks. Suburban networks are moving to 40 nodes or more per square mile. Metro Wi-Fi networks in urban canyons will likely require 100 nodes per square mile.
What about technology improvements? We now have custom metro Wi-Fi infrastructure products. Don’t they work better than they did two years ago? Aren’t some products better than others? Don’t some have more range and coverage? Yes. Some of the new infrastructure products are fantastic. However, all the infrastructure products are operating close to the limits of the FCC regulations already. The limiting factor in terms of range and coverage for metro Wi-Fi networks is the client performance. If a network supports public internet access, it must work with a variety of client devices including low power Wi-Fi clients built into notebook PCs and phones. This can be a challenge and reduces the coverage advantage of one infrastructure product versus another.
Practical considerations also weigh heavily on the node density. Where are the mounting locations that are available? How high are they? They are rarely in the best location for radio propagation. A metro Wi-Fi infrastructure system may be able to cover a square mile with five nodes in ideal conditions (read unrealistic), if the nodes are at the perfect height, in the clear, with no foliage, and no building obstructions. Mount the units lower than required in a cluttered environment, and the same system might need 40 nodes to cover a square mile.
The new marketing number for metro Wi-Fi density is 40 nodes per square mile. Reality is even higher. Cities planning a new wireless networ, should start at 50 nodes per square mile and adjust from there. If the only application is mobile communications for the police outdoors, and every car has a high-powered Wi-Fi adapter with a high-gain antenna, you may reduce the node count. If you live in a flat desert and have access to perfect mounting locations, you may reduce the node count. Otherwise, proceed conservatively. You will end up with a more rational budget, more reasonable expectations, and a better network. Metro Wi-Fi networks aren’t free. But if they are designed properly, they can deliver great value to communities.
Wireless veteran Phil Belanger is co-principal at Novarum, a strategic consulting firm assessing the performance of citywide wireless networks. He and the Novarum team chair the Wireless Networks User Experience Roundtable at the W2i Digital Cities Convention.
back
Related Items:
• Molfetta, Italy
• W2i Finalizes Digital Cities Joint State Briefing Program in Riverside, California
• WiMAX Spotlight Shifts to India
• Novarum Compares Networks, Cuts Through Confusion
• Glenn Strachan, Project Director, Macedonia Connects
• Riverside `08
Comments
Ken Biba
In our experience, if you have a high availability requirement (at least 95% of the locations will have service) then it would be highly unlikely that 5-10 nodes per square mile would work.
At least I have not tested one.
I expect that a minimum for high availability service in an outdoor public safety application is will be closer to 20 per square mile than 10.
That is likely the minimum node density for good service, in an outdoor, high power client deployment, as a rule of thumb.
05:18 PM, 10/02/2007
Cliff Kaiser
If the primary "application is mobile communications for the police outdoors, and every car has a high-powered Wi-Fi adapter", maybe 5-10 nodes per square mile? Maybe even less if the mesh is dynamically expanded by mobile nodes in public safety or public works vehicles deployed as needed similar to Lakewood, NJ Police Dept. Cost of entry is low; ROI quicker. Strategically expand the network for other uses or better coverage as needed.
10:30 PM, 09/27/2007
Post new comment:Only registered users can add comments.
Please Log-in
| 
