Connecting the most unconnected may just be a matter of using what's already there.
The digital divide here in the United States has nothing on India, where a billion-plus residents remain un- or under-served by telecoms. In some large part, this is a feature of the country's growing rift between its relatively small urban upper-crust and rural everyone else. Some 70 percent of the country lives in diffuse villages where electricity itself can be scarce and finding a telephone might involve a day's walk. For ISPs, this does not exactly equal dollar signs.
It's a problem whose solution would seem to intertwined with solutions to the country's more general woes as development, and the opportunities it offers, leave vast reaches of India behind. In lieu of some huge altruistic ISP investment, how can we imagine some of the most undeveloped parts of world finding their way to this very internet?
Animesh Kumar and colleagues at the Indian Institutue of Technology in Bombay have an idea: recapturing unused television "white space" to be repurposed as a medium for providing "backhaul" links in a rural broadband internet scheme, e.g. those segments bridging between major internet backbones and the small would-be subnets at the internet's edge. Their work is described in a paper posted to the arXiv preprint server.
"It appears that rural broadband area is a largely untapped market with great potential," Kumar and co. write. "However, there are significant challenges in providing broadband access in the rural areas, including: (i) small average revenue per user as a fraction of total revenue; (ii) high capital and operation expenditure (including license fees); (iii) affordable backhaul which is exacerbated due to a very large population, (iv) energy cost which is worsened by lack of reliable power supply; and, (v) geographic accessibility issues such as right of way problems."
As the authors note, there is currently a major rural internet initiative underway in India called BharatNet. This is a government-funded program to bring fiber backhaul to village Gram Panchayat, which are local offices of self-governance (sort of like a city hall for clusters of villages). This will help but not alleviate the connectivity problem, Kumar explains, as the internet will effectively dead end in these offices. Gaining access to the internet will then mean getting to these nodes.
What the study authors imagine is a way of cheaply linking these Gram Panchayat PoPs (internet point of presences) with further Wi-Fi subnets in the satellite villages served by those offices: "We envisage that TV white spaces (in the UHF band) can be utilized to backhaul data from village Wi-Fi clusters to the PoP provided by BharatNet."
Key to that envisioning is television white space. This is licensed but unused bandwidth that kind of just sits there. The amount of white space available varies quite a bit country by country, for reasons ranging from reuse schemes to government regulation, but in India it turns out that there is a whole lot for the simple reason that there aren't a whole lot of TV stations. As Kumar and his group explain, the UHF band goes almost entirely unused in India. "It has been shown that in almost all cases at least 12 out of the 15 channels 80 percent are available as TV white space in 100 percent of the areas in India," they write.
Kumar and co. continue: "It is envisaged that a broadband access network can be provided by extending Internet coverage from a rural PoP provided by BharatNet (an optical fiber point), by using TV white space in the UHF band. In such a scenario, broadband base stations operating in the UHF band will provide backhaul from villages to the PoP provided by BharatNet. Each village can be served by an unlicensed-band Wi-Fi cluster. This architecture can be used to provide affordable broadband access-network in (rural) India ..."
The group's proposal is already operating within a testbed connecting 13 villages and hamlets via a white space-based mesh-network linked to a central PoP. It's illustrated below.
Creating something that serves a billion people is another matter, of course. The paper argues that it will take a centralized database of license-exempt bandwidth, to start. The white space-sharing problem here is not between TV providers (primary users) and secondary users, as it might be in the United States, but between different secondary users. Managing these conflicts will take some organization and cooperation.