HP collaborates to identify Wi-Fi dead zones cheaply

The technique enables Wi-Fi architects to test and refine their layouts before a network is deployed.

According to Joshua Robinson, a graduate student at Rice University, there currently is no standard industry practice to identify Wi-Fi dead zones.

“The frequency of dead zones have actually been a huge obstacle to deploying city-wide wireless networks,” he told iTnews.

“Since companies don't advertise how they find dead zones, it's hard to say authoritatively what happens.”

Some providers employ expensive, exhaustive measurement studies that require the network to be tested from every location from which potential users may wish to connect.

Other approaches involve taking a few measurements in an ad hoc fashion and fixing any remaining dead zones after the network is deployed.

According to Robinson, the goal of the new technique is to focus measurement efforts on ‘trouble areas’ that potentially could be dead zones.

The technique identifies locations at which the network should be tested by combining wireless signal models with publicly-available information about basic topography, street locations and land use.

“We develop accurate predictions and use these predictions to avoid spending a lot of measurements in areas that have clearly very good or very poor performance,” he explained.

“This is how we are able to use a small number of measurements to more accurately find a network's performance and identify all the dead zones.”

The research won best-paper honours at the annual MobiCom ’08 wireless conference in San Francisco this month.

By requiring five times fewer measurements when compared with a grid sampling strategy, and ‘far fewer’ measurements than needed for an exhaustive measurement strategy, the new technique could reduce labour and equipment costs, researchers say.

Robinson expects the municipalities, companies, and non-profit organisations looking to deploy city-wide wireless mesh networks to benefit most from the new technique.

He named for example the Technology-for-All (TFA) network that is being built by Rice University in partnership with a local non-profit organisation to wirelessly provide free Internet access to an under-served neighbourhood in Houston, Texas.

“We currently serve around 4000 people,” he told iTnews. “Since we do not have a big budget to test the network, techniques to reduce the cost and time involved in finding our dead zones are very helpful.”

Besides the TFA deployment, the new technique also has been tested on Google’s wireless network in Mountain View, California.

When compared with exhaustive measurement studies of both networks, the technique was found to achieve approximately 90 percent accuracy while requiring less than two percent of the number of measurements performed.

In the short term, Rice University researchers will be focusing on extending their research for use in the network planning process.

HP Labs has a definite commercial interest in the project and has been involved in prior deployments in Taipei. However, no plans for commercialisation of the technique have been announced as yet.

hplabsnetworkingresearchersuniversitywifi