From Technology Review.
Heimerl’s innovation comes in a gray box roughly the size of a microwave oven. It has solar panels on the outside to power cellular equipment inside, along with the software for management functions like billing and analytics. Secure the box somewhere and link it via satellite to a voice-over-IP network, and you’re ready to open shop as a mobile service provider. Heimerl’s nascent company, Endaga, sells it for $10,000
…Just one hitch: it’s illegal. Regional mobile providers hold licenses to the necessary airwaves. Indonesian officials were willing to look the other way, but in general, regulation is a significant hurdle for Heimerl’s vision of universal access. To resolve that issue, he has helped develop a “white space” workaround that occupies unused radio frequencies until another network needs them.
The Endaga company web site does not tell me much.
His research website has more info including some papers:
http://www.cs.berkeley.edu/~kheimerl/
1. http://www.eecs.berkeley.edu/~kheimerl/pubs/vbts_ictd_13.pdf
2. http://www.cs.berkeley.edu/~kheimerl/pubs/vbts_nsdr10.pdf
hth
Conceptually it’s fairly simple: a radio to VOIP gateway. Using satellite for the VOIP side, along with solar power, drops any requirement for fixed infrastructure. I have very little knowledge of cellular radio, but it’s all very proprietary, and the handset’s programming must agree with that of the network. So any frequency hopping would need to implemented on the handset as well, making the customer retail experience trickier. I can’t imagine that any standard handset would suffice without a custom programming story. Not a large hurdle, and simply crossed by retailing your own handsets with the service, for example.
The other part I’m murky on is how the endpoint is established and translated through the VOIP segment. For example, a totally insufficient implementation might use a single VOIP number, and every incoming radio transmission is funneled via the one VOIP channel, and the other end answering their ringing phone would see only the VOIP number and have no way to open a new connection to the calling handset. Essentially you need a full session between the two endpoints. I am sure this is relatively trivial, solved, and implemented by this box. I’m just not sure how it works.
Minor second thought: the term “bridge” might be more appropriate than “gateway”. The exact meanings of these terms are ultimately quite context-dependent, and I’m not familiar with this context. But in my experience, gateways are associated with separate sessions at the boundary (i.e. between radio and VOIP) — like my totally insufficient implementation, whereas a bridge preserves the original session across the boundary. All of the above is nothing more than a hunch.
Gee, maybe if I actually read the article:
> When he left Microsoft and joined UC Berkeley’s Technology and Infrastructure for Emerging Regions program, he encountered OpenBTS, programming code that bridges Internet telephony and cellular phone networks.