In this work we focus on unique Internet connectivity challenges and opportunities in rural areas, especially in developing regions. Providing access in rural areas is particularly challenging due to its unique set of social and technical constraints. In order to quantify the challenges being faced by Internet users in rural areas, we conducted on-site and on-line interviews and analysed network traces from a rural network in Macha, Zambia. Our findings reveal severe local and global connectivity limitations that are unique to rural areas. Local wireless networks experience interference and packet loss due to poor network designs and limitations of WiFi in rural areas. Bandwidth-restricted Internet gateways are constantly congested during normal daytime usage periods. Users describe cost, limited availability and unreliability of Internet access as key barriers to on-line interaction. These obstacles prevent leisurely access to the Internet, including, amongst others, users generating and sharing media. This leads to a more transaction-like or "deliberate interaction" model for many rural users.

Usage analysis reveals unique behaviour for users in rural areas. Most traffic is web-based traffic as opposed to peer-to-peer traffic in developed countries. Social media features even more prominently than in western countries, with Facebook being the most visited website; users are twice as likely to access Facebook than Google search. Further analysis of Facebook traffic shows a high degree of traffic between local users in the village. These unique patterns and constraints form the basis for ICT solutions we propose for rural regions. In order to avoid using congested rural Internet gateways, we propose a set of techniques to localize traffic. VillageShare, a Facebook-based localization application, facilitates file sharing amongst users in the village without the need to send media over the bandwidth-constrained gateway. VillageShare is also capable of time-shifting uploads to off-peak usage periods in order to avoid upload failures. In order to exploit locality of interest in mobile phone usage, we designed VillageCell, an open-source, low-cost pico-cell-like base station, that allows users to make free local cellular calls in a village. It takes advantage of the very high penetration rate of mobile phones in rural Africa, which occurs even though many villages lack cellular coverage. VillageCell is also able to support SMS to instant message client exchange as well as routing calls between VillageCell phones and phones on the public switched telephone network.

The low population density and large village diameters in rural areas of Africa imply a need for novel solutions to spread wireless connectivity to individual homes. Current solutions, based on 802.11 require clear line-of-sight and have limited range. Those based on WiMax are not suitable due to high licence and deployment costs. We propose to use the recently freed TV spectrum bands known as "white spaces", encompassing frequencies from 52MHz to 698MHz, to cover vast distances in rural areas. Our solution, VillageLink, builds on the existing 802.22 white space standard to optimally utilize white space spectrum. We add a feature that allows base stations to allocate optimal channels using inter-cell probing across all available TV channels. Channel probing is critical as frequency selectivity is more dependent on antenna characteristics and non-linearity of RF components in the system than on the free-space propagation laws in the white space band.

In order to evaluate our interventions, we deployed VillageShare and VillageCell in Macha, Zambia and evaluated VillageLink using simulations built on a real 3 km white space link in South Africa. We collected usage logs of these systems both through quantitative and qualitative studies. For qualitative studies we involved users in an iterative design process and made use of on-line interviews to understand user perception of our solutions. Ultimately, we hope that this research will lead to better penetration of wireless networks and improved network performance for users in rural villages, culminating in a more inclusive and representative Internet that truly reflects all languages and cultures in the world. (Abstract shortened by UMI.).