December 4, 2009

Wireless Brownouts?

Are we headed toward wireless data brownouts reminiscent of electricity brownouts?

We’re already seeing wireless networks choke under a surge in demand for data throughput. Cities go dark when demand for electricity exceeds the network’s ability to deliver. What might be the consequences of inadequate wireless network capacity caused by insufficient spectrum allocation?

The public discourse on spectrum policy seems to sidestep the potential for such a data brownout calamity. By discussing tomorrow’s demand in terms of today’s experience we are under-preparing for the inevitable. It is the equivalent of when the telegraph company chose to ignore the transformational impact of the telephone. Only this time it isn’t a single company that stands to be harmed, it is the entire nation.

Yes, we know about AT&T’s 5,000 percent increase in throughput demand over the last dozen quarters. Yes, we know Cisco’s calculation that one smartphone has the data demand of 30 feature phones. Yes, we know wireless carriers will become major distributors of notebook PCs, each of which averages 450 times the data demand of a feature phone (again according to Cisco). But using those statistics to define the spectrum needs of tomorrow is like a general using old tactics to fight a new war.

Telecom innovation has always been driven by three factors: (1) computing power, (2) local storage, and (3) interconnection. Today’s growth in the first two is going to redefine the demands of the third.

The computing power of microchips is forecast to jump 50-100 fold in the next two years. Sometime in the 2012 timeframe new parallel processing microchips will hit the market. Moore’s Law, which many thought might have reached its end as clock speed increases approached the threshold for heat dissipation, now has a new lease on life with single- chip, multi-core parallel computing.

Likewise, storage capacity is soaring and costs are dropping in a manner similar to Moore’s Law. Forecasts envision 1 terabyte storage in the average PC and one-quarter terabyte storage in mobile devices within the next few years.

Increased computing capacity, coupled with increased local storage can mean only that it will be possible to do far more things far faster on a mobile device. This reality, in turn, means far more demand on wireless networks.

Consider just one example of how current trends will affect the wireless network: the convergence of cloud computing with the kind of device power discussed above. Heretofore the combination of computing and communications has produced the capability to go online to access and exchange information in defined sessions. With information stored in the cloud and a powerful mobile computing device “going online” now becomes “continuous networking” where the user’s powerful mobile device is forever connected to and drawing from information in the cloud. Moving from session connectivity to constant connectivity will redefine the mobile broadband demand curve.

The spectrum demand freight train is moving faster than the ability to provide for its rights-of-way. The potential that technology will outpace the ability of wireless networks to deliver is very real. The consequences of such an imbalance are not good for the nation.