dealing with unreliable power in Uganda

March 25th, 2009

Rural Uganda is infamous for unreliable power. Through the month of February 2008, power was on for about 40% of the time in Kabale. We would literally run to the power outlets when power came back so that we could charge our phones and laptops. If we really needed to get work done, we were forced to have an extended lunch at the Litte Ritz, the only restaurant in Kabale that had a generator. And there was usually some contention for the available outlets.

I got a little annoyed by this so I had a fellow volunteer bring me an auto adaptor for my laptop from Canada which would allow me to run the laptop from a car battery.

battery, charger and modem

battery, charger and modem

I was able to rent a battery and home-made charger from Osman, one of the local mobile phone techs, for about 30 USD / month. Osman made me a connector for the laptop adaptor and for my fixed wireless terminal so I could use the internet when power was out. I could get full speed, about 16 Kb/s, when power was out because most of the internet cafes didn’t have generators. Once I had this setup, I was kinda happy when power was out – at least during the day ;)

workspace in Uganda

workspace in Uganda

using wifi routers in Uganda

March 17th, 2009

One thing I noticed about some of the technology I encountered Uganda is that it wasn’t always designed with Ugandans in mind. This makes a certain amount of sense if you consider that there isn’t usually a strong business case to design consumer technology products for regions where low volume sales are expected. One example of such a technology is wifi routers.

Part of my work in Uganda was helping small organizations install shared internet connections using consumer wifi routers like the Linksys WRT54G. The problem with using such a device is that they are designed to a share an Internet connection with modems that have an RJ45 Ethernet connector – so DSL or cable modems – not with the “fixed wireless terminals” provided by UTL and MTN which are used throughout Uganda.

Huawei Wireless Terminal

Huawei Wireless Terminal

These fixed terminal have a serial port which is connected to a computer using a serial to USB converter cable. To use a consumer wifi router with these fixed wireless terminals, you need to connect the modem to a stand-alone computer to share the internet across your organization. This leaves a portion of the wifi router unused, specificity the ‘routing’ or internet sharing capability built into these devices.

This set-up has a number of drawbacks, the biggest being the use of the stand-alone computer. Not only does it require a computer to be on any time someone wants to use the internet causing unnecessary power consumption, the computer is susceptible the following issues: viruses because most organizations use outdated, pirated copies of Windows, hard drive failures because of unreliable power and inexperienced users playing with the operating system’s configuration. All of these things can make a working network / internet connection useless very quickly as I have experienced first hand. And with a lack of expertise in most rural areas, the cost to fix such a set-up is usually significant to the small organizations that I was working with.

Just before I left Uganda, I stumbled upon a way to get around some of these problems. My plan was to make it possible to connect these fixed wireless terminals directly to a wifi router avoiding the stand-alone computer. One way to do this would involve adding a serial port directly to a Linksys WRT54GL or similar OpenWRT compatible router, using a standard serial cable in place of the expensive and often hard to find USB to serial converter cable. Obviously, the firmware would need to be upgraded to OpenWRT and would need to be configured to access the internet through the serial port instead of through the DSL or cable modem Ethernet port.

To make things easier for users, a web configuration page could be added to X-WRT with pre-configured values for MTN and UTL. Since OpenWRT is just a specialized Linux distribution, tweaks like DNS caching, web caching, etc. could be added for slow Internet connections. Another modification would be to convert the unused Ethernet port that is normally used to connect a DSL or cable modem into a regular port that can host another computer. This can be accomplished by changing some OpenWRT configuration files.

The only limitation I can think of is the serial port on the Linksys WRT54GL is limited to 115200 bps. While this was sufficient for the internet speeds in Uganda when I was there last year, finding way to accept higher speed modems should be investigated as things are always improving. Adding a USB port might be an option to deal with this problem but it would increase the price and development time as few (if any?) routers are OpenWRT and USB capable.

Since I’m no longer in Uganda, it’s almost impossible for me to work on this project, however, I thought I would post the idea here so that someone could take off with it if they are interested. If you are in Uganda and are interested in collaborating on this project, please contact me.

things you can’t do in the Cahuita bus terminal

March 9th, 2009
Bus Terminal, Cahuita, Costa Rica

Bus Terminal, Cahuita, Costa Rica

2-in-1 sim card

March 9th, 2009

When I was in Uganda I came across this 2-in-1 sim card at a mobile phone shop in Kampala.

two sim cards with my phone

The device allows you to switch between two sim cards without having to power down your phone and switch them manually. When installed, your phone gets a new menu entry that lets you configure a few settings. The most interesting feature to me was the ability to automatically switch between the sim cards at a fixed interval – for example, every 2 minutes – so you can receive SMS messages and / or phone calls on both lines. Obviously, it’s more useful for SMS messages.

One thing that I found interesting about the installation process was the tool used to cut the sim cards down to a smaller size.

sim card cutting tool

The tool looks home-made to me but I’m not sure if the shop bought the tool from the supplier of the device or made it themselves. Either way I think it’s a cool hack.

punched out sim card

I didn’t run across anybody else with this device in Uganda, however I did meet many people with multiple sim cards. It’s pretty obvious why people didn’t use this device – the cost really doesn’t justify the function. At 12 USD, that’s a third of the price of a cell phone, it’s just not worth it.

However, it’s worth exploring why people have multiple lines. First and for most is cost: if you phone somebody on the same network it’s cheaper. Cell phones themselves help out with this task. When you get a second or third line, you can request to have the same number with a different network code (the first 3 digits of the phone number). If you phone a friend with this specially chosen number on your second line, your address book entry will come up regardless of which line you use. This is good for the person receiving calls but it can cause problems if you try to phone somebody who has two lines. Because only one number is stored in your address book, you may think that somebody is out network coverage or their phone is off rather than just connected to another network. I ran into problems with this when I switched from regularly using MTN to regularly using Celtel (now Zain). When I phoned people using Celtel, my name came up on their phone but the address book entry was not updated. When they tried to call me back, it appeared as if my phone was off or I was out of network coverage – and therefore unreachable – which wasn’t the case.

A simple feature could be added to entry level or emerging market mobile phones to fix this. Any additional phone numbers detected for entries in a user’s address book could be automatically added as a second or third number associated with that entry. When a user tries to phone somebody with multiple phone numbers and the line is not active, the other number or numbers could be automatically tried or at least the user could be notified that the contact they are trying to reach may be using his or her second or third line. It would be nice if something could be worked out for SMS messages as well but that might have to be handled at the network level.

The second reason for having multiple lines – a reason that sometimes gets ignored – is network availability. Outside of Kampala, the mobile phone networks often go down without warning for many hours. I remember one time the MTN network was down for more that 16 hours. If you have a second phone line, you simply switch to a different provider and continue business as usual. Well, except if you need to contact somebody who doesn’t have a second line :-) . The feature enhancement mentioned above could help with this situation if the person you are trying to contact has a second or third line. The enhancement might even encourage more people to get a second or third line to improve the reliability of their communications with other people.

The third reason that people have multiple lines is network coverage. Once you get out of the big cities and towns, coverage is spotty and not consistent across networks. For example, your home village may have coverage with MTN, but the village where you need to attend a funeral may only have coverage with Zain. In this case you merely swap the sim card with the one that is stored between the phone battery and battery cover and continue on.

Another relatively inexpensive feature that could be added to entry level or emerging market mobile phones is space for, or a little holder for a second or third sim card between the battery and the battery cover. I noticed that there wasn’t always space for sim cards in this location or that placing sim cards there would start to wear out the battery cover and it would become loose over time.

I think it’s safe to conclude that having easy – i.e. inexpensive – access to multiple phone lines in Uganda is beneficial to people living there for the reasons that I’ve explained. People who are deriving statistics on mobile phone usage in Uganda should be aware of the regular practice of having multiple phone lines as mobile phone penetration rates would be inflated if statistics are derived solely from customer numbers provided by mobile network operators.

Update: For a more general and academic view of mobile phone use developing countries, check out this post by Richard Heeks.