Friday, 9 December 2011

Understanding the geography and structure of the internet in East Africa

So this is my first proper post in a long time and it is not really about Sudan, so I suppose at some point, I should change the name of my blog.

I am working on a project at the Oxford Internet Institute, along with partners in Nairobi and Butare, looking at how the arrival of four underwater fibre optic cables will change the economic picture of Kenya and Rwanda in the sectors of tea, tourism and BPO (Business Process Outsourcing).

Coming from a social science background, one of the first steps has been figuring out how the internet actually works. This may seem a silly step for someone sitting in an OII seat, but a couple weeks ago I had only a hazy idea in my mind: somehow we were all connected via wires and wireless signals and we used these wires and "beams" to send messages and data across the globe. My colleague Shira added 'When it doesn't work, you turn it off and then back on again. That's how the internet works.'

After a couple weeks of frantic reading, wondering and pestering of colleagues, I can say that this hazy picture is more or less correct. In the simplest terms I can find, I will try explain how the internet works so that I can move past this first step onto greener, more sociological pastures.

More specifically, I will explain how the internet works in East Africa...

First of all (and they should really tell you this on your first day at OII): the internet is not AN inter-net (I have since discovered that the Internet’s capital letter is a bit of a grammatical minefield). The internet is a network of networks. In Arabic, its no 'shabka' but a 'shabakat'. In other words, the internet is plural. Your computer is connected to one network that is connected to many other networks.

And perhaps, more precisely, the internet is a network of intermediated networks. I shall grind my 'it's not globalization because it’s not global' axe here- the internet is not a shared global space, but rather an international space that operates differently in different places. When you connect in one place, your messages are relayed through an infrastructure in a particular geographical and technical space and that intermediation will change your personal experience of 'the internet'.

To put it into the East African context, in the summer of 2009, three underwater fibre optic cables began to 'turn on' faster internet for Kenya and its neighbours. This is a map of all the underwater cables that connect Africa. The three East African cables (SeaCom, Eassy and TEAMs. The fourth LION-2 is still being tested) run along the Eastern coast:

If we think of the internet as the total sum of messages and data, then we can say that East Africa suddenly had access to greater 'bandwidth' (the measurement of data). Before the cables, all messages that came in and out of Kenya had to pass through satellite connections. The reliance on satellites meant that speeds were slow, prices high and bandwidth low. A blogger based in Rwanda provided a nice visual illustration, explaining how the reliance on satellites slowed down his connection:

While internal Rwandan speeds weren't too shabby, international messages had to travel high up into the sky and then back down again in order to reach their destination.

Prior to 2004, Rwanda also lacked an Internet Exchange Point, meaning that even internal messages between Rwandan ISP accounts would have to make that sky high journey. Now Rwanda has RINEX but not all ISPs are connected to it, so many messages must still travel abroad even when they are destined for the house next door. Mark tells me that some Kenyan messages still travel to LINX in London despite Kenya having its own KIXP. In such circumstances, if you want to do something that requires higher bandwidth, you must pay dearly when there are no cables in place.

The arrival of the cables has therefore brought the possibility of much lower costs and higher speeds. However, not all Kenyans or Rwandans were immediately 'switched on'. Rather the cables brought the bandwidth to the port of Mombassa, but its journey inland had yet to come.

(A nice SeaCom video about the laying of the cables can be found here)

Broadly speaking, data can travel wirelessly or it can travel through wires.Wired internet connections might include telephone wires, power lines, fibre optic cables or any other kind of cables, either below ground or underground. Existing telephone wires are made of copper which DSL uses to transmit messages. I have even been told that you can even use powerlines for internet connections. And of course, you can also use an ordinary dial up modem (the kind that makes those satisfying sounds that we all used to know so well. Here, the difference is between dial-up analog and DSL digital). Fibre optic cables are just an improvement on this 'fixed line' technology. Joss, the colleague who sits opposite me said, 'as a basic rule, use cables when possible because they are much faster. When you can't use cables anymore, you use wireless.'

In much of the UK and the US, landlines are pretty well established and so our internet infrastructure has been dominated by cables (most of my friends have a contract with BT, Orange or Virgin involving a modem in their homes. Despite the fact that we spend hours and hours on the phone setting up these connections and often wallowing in confusion for days, we remain loyal to the idea of cables). In other parts of the world, it is much less common for people to have a fixed line telephone and much more common for people to have a mobile telephone. In this situation, wireless internet solutions have taken off.Wireless internet connections might include radio microwaves, local 'wi-fi', GSM/CDMA (these are just competing standards for 3G internet and mobile phone spectrum), WiMax or most recently of all, LTE. Basically, 1G were radio waves, 2G were basic mobile phone communications, 3G allows higher bandwidth mobile phone/internet and 4G has yet to come (although WiMax and LTE claim to already be there). These kinds of technologies are more suited to geographic areas that lack fibre or copper wiring so they are good for Africa, and they are also used in our homes, to connect a laptop to a router.Cost issues are important. Laying down cables is an expensive process with most of the costs (68% for fibre optic) coming upfront. In contrast, wireless technologies can be scaled up, meaning that the upfront costs are lower, but there are higher marginal costs over time. As Joss said, when possible, cables are preferable but financing such projects can be tricky. For this reason, some 90% of Kenyan internet users access the internet through their phone.

Even when the fibre optics are in place, consumers still have to pay for them and of course, those on lower incomes face sizable financial barriers. Richer consumers may very well get a cable connection through a fixed line or pay for fast WiMax, but poorer consumers may rely on their mobile phones. Big infrastructure matters but small infrastructure matters too.

Also, while cables connect whole homes or offices, many mobile based technologies connect individual users, so the distribution of users within a household may be different for this model of internet connectivity.

Furthermore, one's geographical position also plays a role in determining accessibility.

Our project is concerned with Kenya and Rwanda. To give an indication as to how the internet has been brought 'inland', I shall present a few maps.


This map show's Rwanda's national fibre optic grid. This back bone connects different parts of the country to the international cables.

The red dots are connecting nodes or PoPs (Point of Presence). The challenge is to transport bandwidth from these nodes outwards into people's homes and offices.

Kigali City (Rwanda's capital) has deployed a 'WiMax' system in order to provide highspeed internet throughout the capital. A graphical representation is provided below:

RDB, 2010: 16.

This infrastructure has been designed to deliver wireless internet to different parts of Kigali city through a mixture of fibre optic cables (connecting it nationally and internationally) and radio towers (connecting it locally). One WiMax station can cover a very large area, usually a radius of about 50km (or about 30 miles). In addition, one tower can connect with another tower through microwave, so that they can be scaled up relatively easily. Here is a geographical representation of the same network:
RDB, 2010: 15.

So if we again return to the map of Rwanda as a whole and we zoom in one each PoP, we may imagine similar systems taking form in each area. Companies may choose to deploy WiMax or GSM or any other kind of internet connection in each area. In other words, each area will have its own internet infrastructure that reflects the needs (and economic demands) of the area. Below this level, consumers face their own costs and benefits.


My Kenyan map is actually three maps in one. I have overlaid Kenya's terrestrial fibre optic cables with the Safaricom GSM network and current WiMax deployments.

Kenya is a much bigger country and its populations is more dispersed so it's no surprise that its infrastructure tends to be more geographically concentrated (incidentally due to its diminutive size, Mauritius was the first country in the world to deploy Wi-Max throughout its entire country).

While the underwater cables connect Kenya, the country to faster speeds and higher bandwidth, not all parts of Kenya are connected. The pink lines show the current Kenya Data Networks fibre optic cable, the orange shows Safaricom's GSM coverage and the red and blue pegs show where WiMax technology has currently been deployed. Internet penetration is generally restricted to the capital city, Nairobi, its surrounding areas, the road that connects it to the port city of Mombassa and a few other 'hot spots'. In other areas, the government has endeavoured to spread the internet through e-villages and other developmental initiatives. Nevertheless, the internet is a technology that builds on the technologies of the past (telegram, telephone, rail, road) and it has, by no means killed distance. It creates new patterns of proximity and distance and new kinds of intermediations.

Before concluding, I want to add one final point. I have so far shown how the internet is spatially prescribed, but its 'spatiality' is not just restricted to geographic space. Rather, within the technical infrastructure of the internet, space is also expressed through the rules that the internet uses to sort messages and protect users.

Broadly speaking, the internet works through IP (Internet Protocol). Because the internet is a network of networks, it is also a collaborative and interdependent place, where things are both open and vulnerable. To ‘protect’ one’s computer or one’s network from spam or otherwise deviant attacks, users will use ‘firewalls’.

A firewall inspects all incoming traffic and relates it to a series of rules developed by the administrator to protect or censor the user. A firewall can be embedded into a personal computer, an ISP server, a DNS server, or in any other node along the network.

Every physical connection to the internet has an ‘IP address’ that situates that connection in space (both virtually but more often than not, geographically as well). Typically the ‘rules’ that the firewall uses to check incoming messages, look at IP address and compare them to a list of ‘blacklisted’ IP addresses that have been deemed untrustworthy or malicious by other users.

When one region of the world experiences a high number of spam or fraudulent messages from another region, it will ‘blacklist’ the sender’s IP address. Unfortunately this ‘blacklisting’ does not only target the spammer or scammer, but the whole region. In other words, users of the
internet in one area may be penalised for sharing a connection with an ‘offending party’.

(from: Kirstein, 2010. The image and a deeper analysis of the Rwandan case can be found here)

Furthermore, an area may not only be penalised through rules, but also through reputation. The internet opens up the possibility for new representations of place. Many people have never been to Kenya or Rwanda, so their image of the place has been framed by news reports, Hollywood movies and the occasional reference to Obama’s father. Perhaps a Ngugi wa Thiong'o book may surface or a sad film about the Rwandan genocide. These images may not neccesarily instil confidence in a manager looking for call centre workers or for a tourist choosing a destination for his honeymoon.

The internet might allow for Kenyans and Rwandans to adapt these images, projecting new views, new ideas and new relationships. However that opportunity for re-representation is a shared endeavour and has the potential to both improve or damage Kenya or Rwanda’s image abroad. For example, West African ‘419’ email scammers have been described as ‘brand eroders’, as ‘undermining markets’ and as unleashing ‘social damage’ on wider society. The identification of Nigerian IP addresses as sources of such ‘crime’ has resulted in some services blocking all Nigerian traffic from using their sites and has damaged the credibility of ‘legitimate’ Nigerian and African entrepreneurship. West Africans have proposed various solutions to these censures, involving technological innovations (blocking traffic at ISPs) and through the ‘rebranding’ of Nigeria by civil servants and PR agents. Therefore, actors within common places struggle to represent their place in particular ways and to use their different kinds of representation to secure different kinds of extraversion.

For some, representations of instability are beneficial, while for others, stability and governance solicit streams of foreign wealth. This project will therefore treat connectivity both in terms of physical capacities, but also conceptual or social connectivity. How do the cables open up new opportunities for representation of place?

I have tried to keep this post as simple as I can, but I will build on it as I get to grips with the ‘business side’ of things (hopefully future posts will be a bit shorter!).

The main point has been to show how the internet is part of a geographical and technical world. While we tend to view the internet as a ‘global space’ in which all users share a common experience, place still matters and although I have tried to show that the internet DOES open up new possibilities for place representation, it also builds on past distances and dislocations.

p.s. And please correct me if I have gotten anything wrong!


  1. My dearest Laura, welcome to the tech side. Brilliant!



  2. Thanks Olga! You can tell me if I made any mistakes! :)

  3. nice your modesty as always..check out
    cheers and best in oxford - mack

  4. Very interesting post. Thanks!

  5. Thanks Mack! I shall check it out. You keep me up to date. :)

  6. Thanks Gwen! welcome to my blog. I shall try my best to be more up to date. What do you do? Are you interested in the internet in East Africa?

  7. cool post! can't wait to read the next... If I may ask: can I any questions about this subject at some point? just started my internship on a topic that is related to this and being a economic development student, I don't have a clue how these (infra)structures work. I would really appreciate it!

    1. Thijs, thanks for your message. please send me an email,

      I am still trying to figure things out, but maybe we can compare notes. :)