Internet penetration in rural areas, especially in developing countries such as India, is generally poor. Telecom companies do not find it economically viable to deploy wired broadband such as DSL; satellite connectivity is expensive and often slow; dial-up (if available) is always flaky; and cellular data services such as GPRS or EDGE are quite costly to use. Newer technologies for wireless broadband such as WiMax do promise higher bandwidth, but infrastructure costs for deployment in rural areas remain high. How then can Internet connectivity be provided in such areas in a robust and low-cost manner?
One could, of course, ask whether the Internet is required at all in rural areas, given that most of the population is only semi-literate. But this is a flawed argument. The Internet today goes beyond text and includes video, audio, images, and even caters to local vernacular languages. This makes a lot of content relevant to a much larger demographic. Second, the Internet has also become the primary vehicle for delivery of transaction-oriented applications such as buying a train or bus ticket, or making bill payments, and hence is more relevant to rural areas for making it easier to conduct such transactions.
A number of research groups have developed different technologies to provide connectivity in remote rural areas, and I will talk about some of them in subsequent posts. In this post, I will describe an innovative project called VLink done by the Tetherless Computing Group at the University of Waterloo in Canada. The group, led by Prof. Srinivasan Keshav, has been working in this area for the last four years. There were two main principles behind much of the philosophy followed by the group:
- Delay reduces cost: When we normally think about the Internet, we think of it as an always-connected medium where users can surf the web, browse pictures, view videos, etc, all on an instantaneous basis. But this is not always necessary for many applications. For example, if you want to send an email, it is immaterial whether the email gets delivered to the recipient's mailbox instantaneously or a couple of hours later, because the recipient will only see the email the next time that they log onto their computer.
Similarly, if you want to buy a train ticket, it should be sufficient if you were to put in a request right now and receive the ticket the next day when you log on. The drift here is that for many applications, technically called delay-tolerant applications, you need not always be connected to the Internet, and even delayed connectivity is fine to a large extent. The insight of this project is that if users are willing to tolerate some delay, it can significantly reduce the cost of providing Internet access, as I will soon describe.
- Applications and communication medium are separate from each other: Consider various examples of applications such as email, ticket booking, video download, picture download, etc. None of these applications really depend on whether the underlying medium of communication is a satellite link, or a broadband link, or maybe even a dial-up link. And the Internet has been designed in such a manner that different communication technologies can be substituted for each other, but the applications will still work. In technical language we call this layering, meaning that applications work at a higher layer than the communication medium, and as long as the roles and responsibilities of layers are cleanly separated from each other, higher layers won't care about how the lower layers are implemented.
VLink used these two principles to develop a software framework on which delay tolerant applications could be built, and different communication channels could be used in the underlying layers for moving bits to and from the Internet. These communication channels include the following:
- Vehicles to ferry data: Vehicles such as cars and buses that regularly travel between villages and cities are fitted with a wireless router that runs from the vehicle's battery. This router carries a hard disk and supports WiFi. When the vehicle drives past a kiosk in a village, it wirelessly picks up data from the computers in the kiosk, stores this data on the hard disk, and pushes the data off into the Internet when the vehicle drives through a WiFi hotspot that is connected to the Internet. The same process is used to fetch data from the Internet and deliver it to the kiosk.
Even if the vehicle does not stop as it passes a kiosk, the 5-6 seconds of connectivity are sufficient to transfer almost 200MB of data. The system supports all sorts of additional features such as retransmission of data if a vehicle breaks down midway, duplication of data on multiple vehicles to increase the chances of speedy delivery, fragmentation of data so that different parts of the data can be transferred through different vehicles, etc.
- USB keys: Probably the most robust method of data transfer, USB keys are cheap, can provide gigabytes of storage, and can be carried around virtually anywhere humans can go. VLink uses USB keys as a communication medium, and, similar to vehicles carrying data, the data can be split across multiple USB keys, duplicated, retransmitted, and reassembly works regardless of the order in which different data fragments arrive at the destination. In fact, some fragments could be brought in over USB keys while others over vehicles, and everything will still work!
- SMS: Although a very non-intuitive mechanism, it actually makes a lot of sense. VLink can fragment data into 160-byte chunks that can be text-messaged to the destination and then reassembled. This is especially useful for data that needs to be delivered urgently, for example, a high priority email, which cannot wait for somebody to carry a USB key or for a vehicle to drive past. VLink can internally infer that such short high priority messages can be sent over SMS, while larger non-urgent messages can be sent over video.
Robust and Low Cost
Such non-conventional communication mechanisms actually end up being very robust and low-cost. The cost of wireless routers on vehicles gets amortized across all kiosks that a vehicle visits. Similarly, USB keys are cheap and a key can be handed to practically every single person who moves between a village and city. SMS is also available practically everywhere, and in fact many countries such as India have bulk SMS plans to allow the user to send many SMS messages in one single go.
The VLink group has also built some useful applications on this platform. A Flickr application fetches pictures matching a search term, a YouTube application similarly searches and downloads video, an email application delivers email, etc. An open API is also provided, which allows developers to build other innovative delay tolerant applications as well.
Different parts of the VLink system have even been piloted in many locations. A vehicle-based pilot was done in the village of Anandpuram near the city of Vishakapatnam in India, where VLink was used to move e-governance forms for birth certificates, income certificates, and more to the Internet.
A pilot in Ghana helped connect a central hospital in Accra with rural clinics, enabling nurses in remote locations to consult with doctors in the city. And even at Gram Vaani, we plan to use USB keys to ship logs from remote community radio stations to the Internet so that we can analyze them to find bugs, before sending back upgrades and patches to be automatically applied to the systems running at different radio stations.
For more information, you can either write to me because I worked on many parts of the project for my thesis, or you can write to Prof. Keshav, or leave a comment here and we will get back to you.