In this paper we present a peer-to-peer dialup architecture for accelerated “Internet access” in the developing world. Our proposed architecture provides a mechanism for multiplexing the scarce and expensive international Internet bandwidth over higher bandwidth p2p dialup connections within a developing country. Our system combines a number of architectural components, such as incentive-driven p2p data transfer, intelligent connection interleaving and content-prefetching. This paper presents a detailed design, implementation and evaluation of our dialup p2p data transfer architecture inspired by Bittorrent.
I always enjoy the idea of doing more with less and this paper highlights such concept very well. In certain countries outside Internet connectivity is at a premium, and thinking of smart techniques to use it properly is very welcome. As an example, this paper highlights how in countries such as Pakistan it is not so much the “last mile” that is the problem, which can easily provide close to 40 Kbps, but the Internet connection to the outside world, which is limited by ISPs to 10-20 Kbps for each dialup user.
To go around those limitations, the paper presents a P2P dialup architecture for accelerated “Internet access” in developing countries. One can think of such architecture as one that multiplexes scarce and expensive international Internet bandwidth over higher bandwidth p2p dialup connections within the country, building a DialUp BitTorrent system. It is nice to see how peer-to-peer technologies apply to a much wider area of networking than simple illegal file sharing. The idea is that peers offer their resources for others, but in the case of Dittorrent, the resource is not only the downloaded blocks, but also the modem (and thus the connectivity) of the node.
The paper presents an interesting and well-motivated design for a system that seems to solve a real problem. There are several non-trivial innovations that are necessary to adapt BitTorrent to a very different use case. These have been described and evaluated in the paper reasonably well. The paper is missing some important system details, e.g. the expected number of peer nodes and the degree to which nodes are likely to want to download the same file. Also, if the user population is very scarce or diverse, then, this approach may not make a lot of sense. Still, the overall principle of combining caching with aggregating multiple local connections is of interest and will likely come back as the bottleneck moves from the edge back into the network, even in more developed countries...