2010: Guidelines for Successful Large Scale Outdoor Wi-Fi Networks

Novarum has just written an 8-page white paper titled “2010: Guidelines for Successful Large Scale Outdoor Wi-Fi Networks” which contains recommendations to service providers on what makes a successful large-scale outdoor Wi-Fi network today. These recommendations are based upon Novarum’s work between 2006 and 2009 testing over 175 wireless networks in 36 North American cities. Last April 2009, PC World published Novarum’s test results for 3G networks in the US. Earlier in February 2009, Ken Biba, one of Novarum’s founders, wrote an article on MuniWireless about how muni WiFi networks outperformed cellular and WiMAX networks.

According to the Novarum paper, service providers should deploy at least 2×2 MIMO 802.11n for both access and backhaul tiers, 60 nodes per square mile (at least), to ensure the best coverage and performance. Sufficient node density is critical. Measured radio frequency signals from Wi-Fi access points poorly predict performance an coverage:

“More bars does not necessarily mean better service. Rather it is the density of access nodes that is the best predictor of good performance since the uplink from client to base station is the limiting factor in performance, capacity and coverage  . . . The improved coverage of 802.11n does NOT, however, replace node density. In combination with high access node density, 802.11n offers the promise of truly useable and high performance wireless networking with robust coverage . . . Citywide Wi-Fi networks should at the very least meet  - and preferably exceed the performance of cellular.”

Here is a summary of the guidelines for wireless broadband success with Wi-Fi:

  • Modern wireless broadband networks should deliver a useful service with more than 90% coverage for 802.11n laptops at a performance that exceeds cellular data performance expected in 2012.
  • Deploy access network densely – more than 50 nodes per square mile for 2.4 GHz access. Novarum recommends 60 nodes per square mile based on our measurements of over 175 large scale networks.
  • Deploy 802.11n technology for infrastructure – both 2.4 GHz access and 5 GHz backhaul.
  • Different clients will have different coverage and different performance on the same network infrastructure. Base the network design and expectations on the typical client devices to be used in the next five years.
  • Deploy 802.11n technology for clients if at all possible. Avoid 802.11b – these ancient Wi-Fi clients will degrade network performance and every other user’s experience.
  • Expect very good networks to deliver outdoor 802.11n coverage between 90 and 95%.
  • Expect very good (and affordable) 802.11n Wi-Fi networks to deliver outdoor coverage between 50-75% for smartphones.
  • Expect legacy 802.11g based wireless broadband networks to deliver largely unacceptable service of 80% for laptops and below 50% for smartphones.

Download 2010: Guidelines For Successful Large Scale Outdoor Wi-Fi Networks

See also: Novarum Publications


  1. Great research from Novarium, but I have to disagree with the premise that you need to have 60 nodes per square mile. The reason Novarium states that this is necessary is because “the uplink from client to base station is the limiting factor in performance, capacity, and coverage.”

    Wired Towns chose Altai Technologies as their preferred WiFi networking equipment vendor because they — uniquely — have developed several technologies that allow low power clients to communicate back to the base station even in hostile RF environments.

    When I first deployed Altai in Union Square, NY, usage and the network footprint went up ten fold overnight. It’s the same story in Rockefeller Center and in Times Square — very challenging environments, but strong traffic with relatively few access points because the Altai A8 ($8000 list price) has such great range and capacity, can ‘hear’ 100+ low power clients at once over a long distance despite contending networks.

    So yes, generically its true — 802.11n + lots of APs = successful deployments. But Novarium has not tested an Altai network that I know of, and at least for now they have the better mousetrap. Altai is in 50 countries, but only starting to get traction here, and Wired Towns, full disclosure, is trying to help them achieve that.

  2. I would like to echo what Marshall mentioned both from the standpoint of thanking Novarium for their diligent efforts and also cautioning them with respect to the 60 node per mile. In cases where beam forming equipment is deployed, as seen in deployments using Go Networks and Wavion equipment, significantly lower node densities can achieve excellent coverage.

    At the same time, both go Networks and Wavion (at last sighting) were delivering under 20Mbps (aggregate) which is not going to be enough to service the smart phone demand, forget the fixed wireless customers’ requirements.

    I haven’t had an opportunity to play with the Ruckus beam forming equipment yet and am hopeful that we will see similar penetration as was demonstrated by both Wavion and Go Networks with substantially improved data rates.

  3. Esme,

    thank you for sharing. I would tend to agree that 60 nodes per mile is a bit on the high side although it all depends on bandwidth and coverage requirement. It will be interesting to see how these numbers will change when more 802.11n networks will be deployed.

  4. Great information. Today it is more effective to deploy outdoor Wi-Fi using 802.11n that it was years ago with traditional 802.11 a/b/g systems. I’m also glad to see the comment “More bars does not necessarily mean better service” which I’ve been preaching for years. The one thing that most forget to talk about is the amount of wireless interference that can take place.

  5. 60AP’s per square mile could be worst case but I would like to see the details in what you are defining in terms of APs. It would also be interesting to see how you handle backhaul between the APs.

  6. asif ayob says:

    If i wanted to set up a wifi network for a 20mile radius, how much would it typically cost and what equipment would i need.