Thursday, November 25, 2010

Femtocells for Wireless Broadband

Professor Mark Reed, Principal Researcher, NICTA, talked on "Next Generation Wireless: How can we solve the data crunch?" at the Australian National University in Canberra, this morning. This was timely as I was at a workshop with people from the Bangladesh government discussing "e-Government for Developing Nations". One key point is that wireless allows skipping generations of telecommunications technology.

Mark argued that Femtocells can provided greatly expanded broadband wireless and outlined research areas. However, it seemed to me the major issue was how this technology could be integrated into the business models of telecommunications companies (as illustrated by the lack of a business case holding up progress with the National Broadband Network). In other words the research question is: "How do we make money out of this?". As an example, one way would be for the customer who buys a femtocell to share in the revenue from others using the cell (this happens with some public WiFi systems).

Previously I suggested wireless be built into the NBN modems installed in homes. This would provide public femtocells and a very profitable supplement to the wired service.

One technical area for research is how to carry video efficiently. As Mark pointed out video is the major driver for wireless use. But video has very different characteristics to voice transmission and web access. It should be feasible to make video hundreds of times more efficient with a few simple protocol tweaks. Changes to the network topology would make it hundreds of millions times more efficient. Some changes are relatively simple, such as changing packet sizes and priorities, some will require hardware changes, such as putting caching in the cells. While there are many millions of videos which people might watch, there will be a relatively small number which most people will be watching at one time. Also the system can anticipate what people will want to watch and download it when there is spare network capacity.

Another issue is the use of mesh networks. With this arrangement, the consumer's handsets and base stations can communicate with each other, supplementing the fixed infrastructure. This could be used with intelligent and predictive caching.

No comments: