Glenn Fleishman at Wi-Fi Net News seems to have been infected with too much reality and he has a realistic outlook on white space wireless networking.  Fleishman deserves some warning from me that this is the kind of party pooper engineering reality that gets you uninvited from all the cool parties in Silicon Valley because you’re out of step with the tech vogue crowd who go around calling white spaces “Wi-Fi on steroids”.  Life is so much better if we just believe in the magic pixie dust of steroid enhanced white spaces networking even if the networks behave more like roid rage in reality.

There were several great points in Fleishman’s article.  One of the first myths he busts is this notion that white spaces will deliver hundreds of megabits per second of bandwidth.  Fleishman wrote:

“White-space spectrum can only be used in 6 MHz blocks. Even with an extremely efficient encoding, I don’t see how one can get more than 15 to 20 Mbps out of a channel. I’ve seen several statements that white-space networks will hit 400 to 800 Mbps.”

That’s a good estimate, but it’s actually worse than the 15-20 Mbps per 6 MHz block that Fleishman estimates.  Real world white space implementation in Claudville gets 2-6 Mbps which is identical to the Mbps/MHz in Wi-Fi.  That’s not surprising given the fact that these white space networks basically use modified Wi-Fi equipment, and Wi-Fi performance falls in this slower range because it has to live with lower signal levels necessitated by greater coverage and range.

Fleishman also points out the heavily shared nature of white space networking because it operates on unlicensed spectrum.  He wrote:

“You can have distance or speed but not both: the more area you cover, the more users you cover, the more you have contention for air space or time slots, and the less bandwidth available to each user.”

This is also a great point, but it’s even worse than this.  The more area a network covers, the weaker the signal at the outer edges or even in the nearby locations operating behind obstacles or terrain.  Those weak signal zones are severely limited in radio performance and they must operate at the slowest encoding rates.  Furthermore, the weak signal clients have to use up more air time because they take longer to send the same data, and that even slows down the faster clients who have strong signal connections.  So not only are their more contention issues when there are more users in a wireless cell, weaker signals slash the bandwidth pie many times more.

The fact that white spaces operates at the lower frequencies means that the propagation will be better (e.g., lower frequencies can bend around hilly terrain much better than higher frequencies), and this is what people refer to as the “steroid” aspect of white spaces.  But higher propagation is also a liability because it means a much higher probability of collision and radio interference and this is the “roid rage” aspect of white spaces.

This unlicensed aspect of white spaces is a huge problem and it means people can use up all the available spectrum even if they’re going to waste most of it.  Want to set up some wireless cameras and use up all the 6 MHz channels?  No problem, there’s nothing to stop anyone from doing that.  Want to waste white spaces on point-to-point backhauls even though the spectrum is more efficiently used for point-to-multipoint broadband access networks?  No problem, it’s unlicensed spectrum.

Flieshman also points out how WiMax was similarly hyped but that it has now fizzled somewhat.  I would also point out that WiMax at least operates in licensed interference-free spectrum whereas white space devices have to operate in all unlicensed spectrum.  This means that the overly hyped white spaces market is far more likely to fail than WiMax which wasn’t even a high water mark to begin with.  So will white space networks replace cellular and data mobile networks?  Only if you believe in fairies.