Friday, October 30, 2015

MIMO and Small Cells Can Only Do So Much: More Spectrum is Needed

If you assume demand for mobile Internet access bandwidth is going to keep growing in excess of 50 percent a year, as it presently is doing (Cisco predicts bandwidth consumption in Asia will grow at 58 percent annual rates through 2019), then supply also has to be increased.

If you also assume the percentage of higher-performance smartphones will keep growing, then you also must account for much-heavier bandwidth demand. A feature phone tends to use about 22 MB a month. A smartphone tends to use 819 MB each month, while a smartphone used on a 4G network will tend to consume 2,000 MB a month, according to Cisco.

As smartphone adoption grows, so will the percentage of phone customers regularly using mobile data. That also applies to use of smartphones on 4G networks or 3G networks, compared to 2G networks. People consume more data on faster networks

It also follows that significant additional capacity will have to be supplied.


Traditionally, there are three ways to do so. Regulators can allocate more spectrum. Operators can move to smaller cells or use more-efficient antenna technologies, to reach the so-called “Shannon limit (the theoretical maximum efficiency of any communications channel.”

Use of Multiple Input Multiple Output (MIMO) antennas might jokingly be said to prove “Shannon was wrong.” He was not wrong, but the point is that after MIMO--and massive MIMO--are applied, service providers will have wrung as much as possible out of antenna technology.

That will leave new spectrum and smaller cells are the remaining technology tools to boost usable bandwidth (one can think of many ways to provide consumer incentives that reduce demand, but those are not technology tools).

Operators can control network architectures. They cannot control spectrum allocation.



Brookings

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...