While in East Timor I have been brainstorming with the NGO Forum guys on the idea of a city wide mesh Wifi network for Dili. Dili is a reasonably small (10 x 3km) densly populated city of about 700,000 people. It lacks any infrastructure (such as PSTN land lines) that could be used to distribute the Internet.
At present the common approach to broadband Internet is VSAT with a point to multipoint (infrastructure mode APs) Wifi for Internet distribution. There are about 75 VSATs in Dili, and hundreds of towers with grid antennas, Nanostations, and similar devices being used in a criss-cross array of competing networks. This presents a few problems:
However consider a network of 100 mesh nodes. This could be used as a public backbone for IP communications across Dili. It could be a common resource, available for all and owned and maintained for the public good. 100 nodes would mean you are within 600m of a node, so just a simple antenna on a modest, unguyed mast would be required. In many areas a laptop would pick up the wifi signal.
ISPs could then hang their Internet services off a node in the mesh network, and use the mesh to transport Internet to their customers, rather than setting up their own Wifi masts and networks. The mesh network would offer redundancy compared to regular AP type networks, and encourage them to interoperate with other ISPs (for example backing up each others service). Some QoS software could ensure bandwidth across the mesh is shared fairly.
People could also offer other services, for example a local Wikipedia mirror or a web proxy for common content. A network this pervasive and easy to connect to would be great for VOIP – just plug in a SIP phone or use a Mesh Potato for free local calls (or mesh telephony). The magic of mesh networks means that as more people join the mesh it’s coverage improves.
One hundred mesh nodes would not be expensive, for example even with a good quality router in a weatherproof case, 8 dBi antenna, 3m mast and installation by trained locals a budget of $200/node would be quite possible. Thats a total of about $20k, or about the same price as two VSAT terminals. Creating and installing the network would create much needed local employment – as well as installation work the antennas could probably be fabricated locally.