These unused channels, if they can be efficiently and accurately identified, promise a big boost in spectrum for Wi-Fi use. The biggest demand for that spectrum is indoors, according to Chandra. A six-month analysis of white space spectrum in Hong Kong locations found that there is about 40% more white space spectrum available indoors than outdoors.
The problem, according to Chandra, is that "most trials and studies of white spaces done before have focused on outdoor scenarios."
Microsoft Research has been working with CUHK academics since 2010 on these issues, including research aimed at showing how to use this extra spectrum.
The FCC allows two methods for detecting free channels: spectrum sensing, by the device radio, or querying an Internet-based geo-location Web service. According to the WISER research paper, the most commonly used method is the geo-location database, in part because spectrum sensing is costly; and at low thresholds, it's difficult to do accurately with off-the-shelf hardware.
The geo-location method doesn't need hardware and it's easier to deploy. But it has "inherent inefficiency," according to the researchers. That's because it uses propagation modeling rather than direct measurements to identify available spectrum "and hence, is very conservative in the channels it returns for a given location."
In July 2013, the FCC approved for use a white spaces database created by Google, designed to track vacant TV spectrum.
In essence, WISER is designed to draw on the strengths of both approaches and sidestep their weaknesses. Low-cost spectrum sensors make for accurate identification of indoor channels and to do so cost-effectively; the use of a local geo-location database relieves the clients from having to do their own scanning.
WISER was designed to minimize the cost of the spectrum sensors, and to improve the accuracy of identifying usable white space channels, without causing interference itself. A prototype WISER network was deployed at Ho Sin Hang Engineering building on the CUHK campus.
Among other conclusions, the prototype deployment found that "it suffices to identify strong channels via long-time sensing and then focus resources to track the slow-varying white space availability of weak-to-normal channels." The study also found that for a given white space channel there is a "strong correlation in signal strengths and white space availability across different locations. This suggests that we can infer the channel vacancies of multiple correlated locations from those of one or a few representative locations."
The research also found that the "indoor white spaces have different characteristics from the outdoor ones. For example, there are more contiguous unutilized TV channels indoors, which are able to support high bandwidth communication."
Sign up for Computerworld eNewsletters.