Friday, April 20, 2012

Fuzzy Thinking on Wireless "Spectrum Crisis"

One apparent problem with many discussions of how much bandwidth mobile service providers will need in the future is that some observers confuse tactical buying and selling of some blocks of spectrum with the overall bandwidth trend; and today's problems with tomorrow's problems. 


"The wireless carriers say that in the next few years they may not have enough of it to meet the exploding demands for mobile data," writes Brian X. Chen in The New York Times. But some think spectrum crisis is imaginary, and point to some specific proposals, as a commitment by Verizon to sell some spectrum if it is allowed to complete its purchase of AWS spectrum from Comcast, Cox Communications, Time Warner Cable and BrightHouse Networks. 





Other say there obviously will be a spectrum shortage if demand keeps growing 60 percent to 100 percent a year. It is true that technology can help. Better signal compression, more efficient coding, Wi-Fi offload and retail financial incentives can help. 


But is is much more difficult to create usable mobile bandwidth than it is to create additional fixed network capacity. That's mostly a matter of physics. To avoid interference, discrete blocks of frequency are used by only one service provider, though those firms can re-use the same frequency across nearby cells, by using frequency planning.


For fixed network providers, each discrete provider can, in principle, use a full range of frequencies, even when other licensees have exclusive rights to use discrete frequencies in the wireless domain. U.S. wireless service providers have the use of frequencies in the 700 MHz, 800 MHz and 2 GHz bands, for example. All others generally are reserved for other users. And even if higher frequency spectrum were available above 2 GHz, it is, for reasons of physics, difficult to economically use those frequencies in a mobile communications context.


Fixed line providers can use much more spectrum. Where mobile service providers are restricted to specific frequencies in three main bands, each separate cable operator can use all frequencies between about 1.6 MHz up to about 1 GHZ, the real limitation being the cost and signal propagation characteristics, not the ability to use frequencies above about 750 MHz. 


By way of comparison, then, where a mobile service provider might use paired 10-MHz channels, representing 20 MHz of total capacity, and where one single mobile provider can use those specific frequencies nationally, each cable operator can use more than 700 MHz. In other words, every cable operator has an order of magnitude more "raw bandwidth" to work with, compared to any mobile service provider. 


To be sure, many mobile operators run two or three discrete networks, each with 10 MHz to 20 MHz of spectrum, there are clear physical limits, compared with any cable network. 


Landline telcos generally use optical fiber, and different signaling methods, that support hundreds of megabits per second to gigabits per second for every user, in principle. Consistent with the business case, landline telcos also could use multiple discrete wavelengths to reuse all their frequencies over and over again. 


That means it always will be easier for fixed networks to re-use frequencies that, in the wireless domain, must be reserved for use by only a single provider.


Also, most spectrum usable for mobile communications already is licensed to "somebody else." There is very little "fallow" spectrum.  The point is simply that there is no comparison between fixed network potential bandwidth, and mobile potential bandwidth.


Observers who argue there is no need for more spectrum sometimes argue that "new technology" can fix any mobile service provider capacity problems. New technology can help. Alternative technology can help. Creating incentives for people to use finite spectrum wisely can help. 


But there are clear indications that mobile network physical capacity will not be sufficient for future needs. Denying that strikes some of us as fuzzy thinking. 












No comments:

Many Winners and Losers from Generative AI

Perhaps there is no contradiction between low historical total factor annual productivity gains and high expected generative artificial inte...