Two weeks ago I wrote about how the heterogeneous network model is replacing the cellular model of mobile phones coverage areas.
After writing that article, I visited some friends living outside DC. They told me how their house has very poor mobile phone coverage because it’s in a low-lying area. The phone company’s solution was to install a femto-cell in their house that connects to the network via their Internet service. The HetNet is real.
At first I thought of this as just an issue of efficient frequency reuse to minimize unnecessary base stations. It turns out that mobile data service is approaching what could be called a crisis. The wireless industry believes wireless internet access has become “viral” and will grow exponentially. They predict wireless traffic will grow 25 to 30 times by 2015 and 1000 times by 2020. In some ways this is analogous to the the growth of the internet in the 90s. The industry has motivations to overstate growth, but experts believe the predictions are reasonable. 
There are three ways to support increases in wireless data:
- Increase spectral efficiency - Shannon’s Law defines a theoretical maximum data rate that can be sent within a given bandwidth at a given signal to noise ratio. Current mobile systems operate close to this limit. Not much of the projected increase in traffic will come through improved spectral efficiency.
- Increase the spectrum allocated to mobile data - Around 300 MHz of spectrum is allocated to mobile data in most regions. The industry has been lobbying for more spectrum since the 80s because this is the cheapest avenue to support more users. The industry predicts spectrum allocated for mobile phone will increase threefold.
- Spatial Reuse / “Smaller cells” - Most of the increase will have to come having base stations with smaller coverage areas.
As frequencies above 10GHz become usable, I suspect small-cell mobile traffic will move there, providing more than the predicted threefold increase. People wonder if this spectrum will be useful because of the higher free-space path loss and greater reflection and attenuation by objects. These are the exact same reasons I heard in the 80s in Popular Communications Magazine for why it would be difficult to use frequencies above 1GHz. Today we use 5.5GHz Wi-Fi without a second thought.
Dynamic spectrum access may provide more spectrum, but it won’t help in the most densely populated areas.
I predict mobile data traffic will increase 1000 times, and in urban areas much of the in increase will come from spectrum above 10GHz. The extent to which mobile data is a bottleneck will define how mobile devices evolve over the next ten years.
Further Reading
Riding the Data Tsunami in the Cloud by Jens Zander and Petri Mahonen - This is what convinced me the problem of mobile data will approach a crisis. It is also the source for the image.