DOT-COMments: Deploying Low-Energy ICT A technical overview

Deploying Low-Energy ICT A technical overview

Dot-EDU recently set up a teacher training computer center in rural Uganda, and a brief article appeared in the December DOT-COMments, Low-energy Internet for Education Where Electricity is a Challenge. Many people contacted the deployment team for more information on the specific technology that we used, and we thought it might be helpful to share some ways in which this effort could be repeated--a sort of technical overview.

For those who did not see the article, dot-EDU is attempting to solve a common problem for rural technology labs. The quality of electricity in these outlying areas can be poor (frequent power cuts, brownouts, surges), and standard equipment does not survive well. Even ordinary uninterruptible power supplies (UPSs) wear out quickly. Of course there is the further problem that the lab often remains unusable during the day when power cuts happen.

Technical Needs
We set up a low-power lab that runs primarily on 12 volts. A standard charger/inverter receives power from the mains and charges a set of sealed gel batteries. The lab runs on these batteries, and can continue running for about six hours without AC power. In addition, the charger/inverter unit can take the 12 volt power and convert it into AC power for equipment that cannot run on 12 volts.

In this case, it is a thin-client lab, meaning that a set of terminals connects to a server (just an ordinary desktop PC with server software installed). The thin clients have no moving parts and consume very little power. The desktops are outfitted with LCD monitors, which also consume very little power. For individual users sitting at a terminal, the experience is almost indistinguishable from sitting in front of their own computer; the only difference is that all the processing is done on a server. This allows the lab to consume less power, and it means that only one computer needs to be updated for virus protection and security patches, etc.

The inverter/charger that we used is a Tripp Lite APSX1250 (available in the US for about USD 400). This single unit can receive 240 volt AC power, charge a set of batteries, and output conditioned 240 volt AC power. Tripp Lite also makes similar units for countries where 120 VAC power is used. For smaller labs, it would be possible to use a Tripp Lite APSX750 (available in the US for about USD 300). Batteries are available in many varieties. Sealed maintenance-free batteries are perhaps best, although lead-acid batteries can cost less. Sealed 12V batteries are available in the US for about USD 1.25-1.70 per amp-hour. In other words, a 100 Ah battery should cost $125-175, depending on quality and brand. The number of batteries needed, of course, will vary based on how much equipment is in the lab. Our setup in rural Uganda used 1200 Ah, which was much more power than we actually needed, but we wanted to provide for future expansion.

A small lab of five terminals, a server, and a small printer should be fine with 600 Ah or so. That smaller lab could also use the smaller Tripp Lite inverter/charger (APSX750). This would put power equipment costs at about USD 1000 or less.

The terminals we used (Wyse 1200LE) cost USD 300 each. For about USD 400 each, it is possible to choose a terminal that also offers support for USB flash drives, floppy drives, etc. In our scenario, all storage and media are on the server.

We chose to use Windows Server 2003 Terminal Services, because setup and lab manager training time are minimal. Moreover, running standard Windows applications may be helpful for training. That said, software becomes a significant expense. For a five-terminal lab, the cost for the OS would be about USD 1600, although this could be less through academic discounts and in-country purchase.

Linux would offer, we believe, a strong option at larger scale. This user experience would be similar to Windows, there is little licensing cost, and management of many sites could be easier using Linux. However, the configuration time would be higher.

If one were going to build several labs, we would recommend considering the idea of placing servers in a central location and then placing only terminals in the labs. In places where Internet connectivity is good (fixed wireless is often a good option), this could make sense. The server could be located in a major city, with staff to maintain it. Each lab would have very, very minimal power requirements, and it would be possible to deploy new labs quickly.

Thin Client is the way to go
There are several reasons why we believe the thin-client model is worth considering.

First, the initial hardware costs are lower, especially at larger scales.
Second, the power requirements are much lower, making it possible to build solar-powered or battery-powered labs.
Third, by having fewer machines to maintain, we believe that viruses and other problems can be minimized.
Fourth, the individual terminals have no moving parts so they should last longer than conventional PCs.
Fifth, it is easy to lock down the terminal users to particular applications (web browsing and word processing, for example) to prevent the use of the computers for games or other non-essential activities.
Finally, there is less temptation for theft of these devices, because they are not useful unless connected to a server.

Our prototype effort suggests that this can work. Our belief is that this technology should be considered in projects where the goal is to set up many sites in difficult conditions.

For More Information, Contact:
Partner: Scott Gunn
Consultant to dot-EDU, Education Development Center
Tel: +1 508 720 1500
Email:

Related DOT-COM Activity
Uganda - Piloting Low-Energy, Low Cost Thin-Client Appliances to Link Teacher College and Primary Sc

Related DOT-COMments Newsletter Articles

Low-energy Internet for Education Where Electricity is a Challenge From Issue 14, December 2005