Part 2 of write-up on ultra-low-cost lightning detection network. See Part 1 for background.
[By Jakke Mäkelä and Niko Porjo]
This part summarizes the cost and business case estimates made for the project. The analysis suggests that an extremely low cost might be possible, making the solution suitable for use even in developing countries like Sri Lanka. However, we could not find a way to motivate anyone to fund the R&D part. Thus, we are not pursuing this further for the time being.
Business case: both hardware cost and data transmission costs are kept so low that building and maintaining network is realistic to perform as a public service. Data transfer can be made over mobile phone network (SMS, GPRS, 3G…) or landline if available. Multiple operators are made to compete to keep data costs down. Hazard indication to end users will need to be wireless to achieve real-time warning.
•An exact business case is difficult to determine, as it is for any lightning warning system.
•Situation in Sri Lanka: Tens of deaths reported annually, real number of deaths and injuries unknown. Commercial detection systems too expensive to maintain.
•Absolute worst-case scenario: If SMS sent less than once per min, reliability of network becomes poor. Thus cannot transfer much less than this. During storms, whole network would need to be transmitting 60*60=3600 SMS/hour. In Finland, cost of SMS on one operator is ~5 cent, so cost would be 180 EUR/hr for whole network. Assuming 6 hours/ day of storms in active seasons, would mean 30 kEUR/month, which is of course completely unacceptable. But in practice, there are fixed-cost deals available from operators. Multiple operators are absolutely needed in order to maximize price competition and minimize risk of monopoly pricing. But there are SMS-based systems in existence which are affordable (for example Nokia Life Tools http://en.wikipedia.org/wiki/Nokia_Life_Tools) which means that costs could be kept reasonable if there is political push.
•Transfer rate over GPRS, if available, is even in extreme case ~160 B/min or 1kB/hour per station. For total network, transfer is ~60 kB/hour. Per month, amounts to ~100 MB. On one Finnish operator, GPRS cost can be ~1.5 EUR/MB, so cost would be <200 EUR/month. Clearly GPRS would be the preferable channel where available.
•Sensors need to be stockpiled to allow them to be replaced quickly if needed, so at least 100 need to be built. Smaller calibration networks can also be developed in parallel and run in a suitable country.
•Main components are radio receiver, GPS clock, GSM/GPRS, processing unit. Battery may be most expensive individual component. ideally will run on AC power, but need to have backup battery/UPS capable of multi-hour operation in case of power failure. Unit cost of 400EUR should definitely be reachable (cost of full network 40kEUR), though profit margin to manufacturer is then low. This part may need to be subsidized.
•Setting up of network is low-cost since all sensors are autonomous and operate by wireless network. Slow deployment is possible.
•Central unit can be a tabletop PC. Redundancy and power supply needs will increase cost significantly, but main algorithm is simple.
•Operating costs can be minimal if GPRS can be used.
•Cost of transmitting hazard information to users via Cell Broadcast is largest open question.
Funding and implementation
•No funding has been found.
•The organization Geoscientists Without Borders has been funding projects which are similar to the proposal: http://www.seg.org/web/foundation/programs/geoscientists-without-borders
•The target of the pilot is to improve local R&D competence in Sri Lanka, but kicking off the project could require an investment that is difficult to find locally. National foreign aid organizations (Finland or Sweden) might be approached for projects of this type, especially if some of the testing can be done in Europe (enhancing local knowledge also).
•Lots of small semi-official warning systems are known to exist, but limited info in public domain. Data transfer is almost always by fixed-line Internet, which can be unreliable especially in developing countries. Mobile wireless networks have better reliability (though not perfect).
We continue to think that the idea would work in principle, but there is no real way to make it successful commercially. We need to feed our families, and cannot do it.
If someone is interested in making this a non-profit open-source project, the crucial documentation is already in the public domain and just needs to be collated together. There are some major engineering issues to be solved, but if profitability is not a requirement, they are likely to be solvable.