10-Gbps Wireless Access in 25 Years?

Observers disagree about how much bandwidth people will need, and how much will be available for sale, in various markets in the future. But some observers think the typical U.S. end user will have a breath-taking amount of access bandwidth.

Mobile devices will have the power of a supercomputer, argues Donald Newell, AMD Server CTO. To be more precise, a then-current smart phone will have more processing power than today’s servers, Newell argues. That isn’t even the most-surprising prediction. More startling, in all likelihood, is the notion that a typical wireless consumer will have access to 10 Gbps.

Cisco thinks terabytes is a reasonable expectation for U.S. consumers, in a couple of decades. Cisco's Dave Evans, Chief Futurist, thinks at-home consumers will have access to Evans says multi-terabit Internet connections.. "I could have an 8-terabit per second connection to my home," he says. "That's more connectivity to my home than most countries have."

As a result, the core networks will operate at petabit per second speeds, about 10 to the 15th power, about three orders of magnitude bigger than terabit networking," Evans says.

Phones will have more than a terabyte of local memory," adds Mark Lewis, chief strategy officer at EMC, who predicts that all of our digital information will be backed up over the cloud. "If I lose my phone, I can pick up a new one, enter my code word, and it will re-identify me and push all of my information out to my new device."

For wireless networks, typical speeds will be as high as 10 gigabits per second, as fast as the fastest optical core networks today, some would argue.

Bandwidth increases on that order of magnitude, at least in the wireless arena, will require more than spectrum allocation. It will require continued significant advances in signal coding and compression, with some likely changes in network architecture as well. Additional spectrum will help, but it is hard to see typical mobile users getting 10 Gbps without robust new developments in coding.

If not, using today’s technology, cell sites would be so small they would be virtually indistinguishable from a fixed connection, in which cases “mobility” would not be possible.

But 25-year horizons are not meaningful, and predictions for what the world will be like that far out almost always are incorrect. One might find more success betting against today’s 25-year predictions instead.

That is not to say Moore’s Law is repealed, or that users will stop demanding more bandwidth. It’s just that linear projections almost always are wrong, over the long term.

On a relatively immediate basis, though, some projections that can seem outlandish are directionally valid enough to support rational business planning. Netflix, for example, has supported its business by mailing DVDs to customers. It began doing so because there was at one point no way to support delivery over the Internet, even though its very name suggests that possibility.

Netflix CEO Reed Hastings claims that back when even cable modems and digital subscriber line were not available, “we took out our spreadsheets and we figured we’d get 14 Mbps to the home by 2012, which turns out is about what we will get.”

“If you drag it out to 2021, we will all have a gigabit to the home,” Hastings argues.

Still, Netflix took the rational route and did not build its revenue model on bandwidth that wasn’t available; it built on what was feasible at the time. Lots of application service providers based their businesses on inadequate bandwidth and server infrastructure in the early 2000s, and most failed because of those assumptions.

Now, lots of providers are about to make a business out of cloud computing, which is the same concept, but in an infrastructure environment that has changed dramatically.

Timing might not be everything, but it is close. For that reason, no rational executive can build a business today based on expectations of 10 Gbps consumer mobile connections. But the direction is clear enough.