Dili Village Telco Part 1

We hopped off the plane at Dili airport after a short 1 hour flight from Darwin. It was just after sun up but the tropical humidity and heat hits you right away. Darwin has a similar climate, but no sign of manicured lawns and carefully detailed tropical gardens here.

A large sign at the airport says “Goodbye conflict, welcome development”! Not exactly reassuring but I have to say it’s good to be back.

Lemi Soares, the head of the Timorese side of the project, met us at the airport. Despite our best efforts to saunter past customs they pulled us up and we spent 20 minutes itemising the 70kg worth of Village Telco kit we brought with us. The customs people were actually pretty cool and friendly, genuinely interested in the project. A mere $125 in duties (2.5%) and we were out of there.

As a concession to Rosemary the hotel is step up from our last visit. This year we have heated water for showers, and a toilet that works. Last year we tossed ladles of cold water over us to shower, and often used the same ladle to flush the toilet. The air conditioner is semi-functional as usual but still a big relief from the afternoon heat and humidity. And very cool – the hotel has it’s own VSAT for Internet! But it feels like I am cheating this year, somehow less genuine. We Westerners are pretty soft. However I can recommend Hotel Vila Verde if you would like a nice place to stay in Dili.

So here we are back in Dili, Timor Leste, with a plan to install a 100 node Village Telco network. On this trip we will train the Timorese side of our team, who are from an organisation called Fongtil. We will then work with the Fongtil guys to install the first 10 nodes. The Fongtil guys will then install the remaining 90 nodes over the next few months.

Timorese like Lemi are veterans of many well meaning technology projects that break a few weeks after the foreign people who installed them leave. Support strategies like remote access, IM, and VoIP are near impossible due to limited Internet bandwidth.

For example at Fongtil HQ I have 2 second ping-times to Australia (which is only 500km away), and this page took 5 minutes to render. Fongtil has a VSAT link that is being shared amongst too many people. Email works OK, however Lemi made the a good point, “Many problems are impossible to solve via email – you need some one in front of you”.

In the first world we rely on good connectivity and nearby geeks to solve our Linux or Asterisk configuration problems. Especially when getting started in a new technology. Lack of skills, language issues, and a lack of connectivity can be formidable challenges.

A good example today was flashing a Nanostation 2 with Supernode firmware. Lemi spent three evenings last week trying to get this to work before I arrived. I managed to get it to work in 5 minutes this morning. However when I tried to repeat it for Lemi, I hit the same problem he did and we were stuck for 90 minutes! Getting anything done is just more difficult in the developing world. Anyway I will update the Dili VT Wiki to make sure we don’t hit the same problem again.

Making complex systems like the Village Telco to work here is a big challenge. But that is what we must do if the Village Telco is going to live up to it’s promise. Getting it working initially is part of the problem. Keeping it working without Wifi gurus baby-sitting it 24/7 is the big challenge.

Lemi and the Fongtil team have some experience in point-point Wifi
links across Dili, however reliability has been a problem. For example trained personal often leave development projects to seek more secure of higher paying jobs. A typical project may rely on just one person for it’s up time. If that person is away or unavailable, networks fall over. Another issue in Dili is the interference from the many point-point Wifi links criss crossing the city. Without adequate copper Wifi is the only way to move data across town here.

Then there are cultural issues that can be very different to Western thinking. These can clash with the maintenance needs of high tech software/hardware systems. I have some specific examples but I can’t think of a good way to express them on a public blog post without appearing condescending. Nevertheless, cross-cultural “wet ware” issues are just as important as the hardware and software.

So I am working on some strategies to make the Dili Village Telco work, and in particular to keep it working when I head back to the land of low humidity:

  1. Make sure everything works well before you leave the 1st world. Get all the bugs fixed at home, rather than wait until you get here to “finish things off”. So Rosemary and I spent the last 4 weeks assembling the network at our house in suburban Adelaide. Mesh Potatoes, servers, Supernodes, Wifi APs. All tested and configured, and the Dili network design documented.
  2. Bring in everything you might need. You would be amazed how much time can can be wasted due to a bad Ethernet cable or a missing tool you thought you could buy locally.
  3. Make the Village Telco really easy to set up. I think we are close to that with Mesh Potato configuration. Still a lot of work needed on the server side, and we need some tools/techniques to make Wifi networks easy to install and maintain.
  4. Provide a fall back path should a box stop working. The Fongtil guys must be able to recover from any problem without needing me or any other 1st world support. Best recovery technique I can think of is to re-flash a Mesh Potato. Bring it back to known defaults. Much easier to have a known starting point then debug arbitrary conf file problems over email.
  5. Spend most of the training time on showing people how to fix bugs. I intend to break the network in various ways, then stand back while the Timorese guys try to fix it. Much more important than stretching the envelope on the technology.
  6. Make sure multiple people are trained. Encourage them to train others. Make Mesh Potato configuration as simple as possible so training needs are trivial.
  7. Make the Village Telco redundant with various fall back modes of operation. For example if the server fails the Mesh Potatoes should still be able to make local calls.
  8. Budget much more time than you normally would. Life just moves more slowly in the developing world.

Lemi has done a good job promoting the project locally. Last week he presented the project (with a cool Mesh Potato demo) to a local government conference. Some people from regional governments already want 10 in their districts! There are some really cool local call applications. Many of the GSM calls here between are over distances of just a few 100m. In some cases people use GSM handsets to call the next room! Another idea is linking adjacent police stations over distances of a few km in country areas where there is no GSM coverage.

Curiously, local calls are the killer app here. Even gatewaying to GSM seems a secondary concern. Many people here call the same number several times a day, e.g. intergovernmental departments, or even churches calling each other. International VOIP for the general population is a non starter as there is no bandwidth out of the country. This is a paradigm shift for me as our recent Village Telco focus has been on “ease of use” when connecting Mesh Potatoes to the Internet.

What matters here is Mesh Potato set up and reliable, long term operation. A GSM handset works here. Every time. That is the quality bar for telephony.

So first this system must be shown to work. The Timorese government are wary of “use once” foreign technology. First training course on Monday – where we show people how to install and configure Mesh Potatoes.

Links

Dili Village Telco Wiki

Some posts from the 2009 trips to East Timor:

East Timor – OLPC Pilot 2
Metro Mesh Networks for East Timor
East Timor – We Really Do have Road Rules Now!

4 thoughts on “Dili Village Telco Part 1”

  1. It’s excellent to be here. Local people wonderfully welcoming and really impressed with turn out and application of Village Telco attendees. They have mountains of enthusiasm and pro-activity towards project.

Comments are closed.