ICT4B - A Scalable Enabling IT Infrastructure for Developing Regions

There are thousands of stand-alone projects that aim to bring information and communication technology (ICT) to developing regions, but nearly all depend on existing hardware and infrastructure developed for affluent regions. These imported technologies fail to address key challenges in cost, deployment, power, and support for semi-literate users. This proposal develops the key technologies and infrastructure to enable these projects, and many new applications that were previously intractable.
We address these challenges with novel technology, and validate impact via two real-world pilot deployments in two developing regions. Our goal is not just to understand two specific ICT applications, but also to demonstrate that the underlying ICT architecture and technologies truly help developing regions. To this end, this proposal includes two faculty from social sciences to ensure that the work enables real-world solutions. Supporting partners (Intel, HP, IIT Delhi, Markle, Grameen Bank) will provide project guidance and on-site help. IIT Delhi and UCB will co-teach classes on IT for developing regions and encourage student exchange to facilitate research.
The technology strategy is to attack the key challenges of cost, power, deployment, support, and literacy. ICT4B will provide order-of-magnitude improvements in device cost, infrastructure and networking cost, and power consumption. Key deliverables include:
1) novel low-cost, low-power devices;
2) a new approach to low-cost networking based on intermittent connectivity (rather than persistent connectivity as in the Internet);
3) a userinterface toolkit to support poor literacy via novel
speech recognition, and a variety of sensors for environment and health applications; and
4) a three-tier architecture with proxies and data centers to support low-cost devices with more functionality, easier development, over-the-wire reprogramming, and usage monitoring for social science research.
The 10-100 times reduction in device cost stems from co-design of devices and infrastructure, system-on-a-chip integration, and use of open standards. The 10-100 times reduction in device power usage stems from low-power circuits, using less CPU power due to help from the infrastructure, and low-power intermittent networking. The 10-100 times reduction in infrastructure cost comes from intermittent networking, extensive sharing, and novel architectural approaches to user and system support in the field.