The following Case Study of Sub-Saharan Africa was put together following the "2nd IHY-Africa Workshop" 11-16 November 2007, Addis Ababa, Ethiopia,
and the "Internet & Grids in Africa: An Asset for African Scientists for the Benefit of African Society", 10-12 December 2007, Montpellier, France. It contains most of the information presented plus much more background information.
Internet coverage and performance in the Sub Sahara region of Africa (see Figure 1, for the UN definition of the Sub-Sahara) is exceedingly poor. This is illustrated by the number of Internet connections per city worldwide seen in Figure 2 where it is seen that Africa stands out as almost lacking any (i.e. it is dark). Figure 3 illustrates how Africa has very little submarine cable connectivity (white lines) compared to the rest of the world.
Figure 1: UN Definition of Sub-Sahara |
Figure 2: World Internet Connection Density |
Figure 3: Submarine Cables 2007, from telegeography.com |
|
|
|
There are many reasons for this poor connectivity including: electricity (see Figure 5 for the light at night for the world and compare Africa with the rest of the world), import duties, skills (see Figure 6 for a cartogram (where territories are re-sized according to the subject of interest - all cartograms shown here are from http://www.worldmapper.org/) of tertiary education spending growth), disease (Figure 7), protectionist policies, corruption (see Figure 8), conflict (see Figure 9).
Figure 5: World light at night |
Figure 6: |
Figure 7: HIV Prevalence |
Figure 8: Corruption Perception Index |
Figure 9: |
|---|---|---|---|---|
|
|
|
|
|
Another way of looking at the poor situation in Africa is to look at Figure 10, illustrating the lack of Internet users in Africa compared to the rest of the world. At the bottom left of figure 10c is shown the growth of Internet users and cell phone subscribers. This may suggest that cell phone infrastructure may be a very valuable way to leverage Internet growth.
Figure 10a: |
Figure 10b: |
Figure 10c: |
|---|---|---|
|
|
|
International capacity to African countries is mainly provided via satellite or via fibre links. Satellite links are not only much more expensive (300-1000 times) in terms of $/Mbps but they also induce long delays of over 400mseconds that result in lack of interactivity and poor performance. However in 2004 only 14 of 49 sub-Saharan countries had access to fibre according to NEPAD. In fact as seen in Figure 11:
Figure 11: Fibre links to and within Africa and the locations of universities |
Figure 12: |
|---|---|
|
|
There is only one large-scale intercontinental fibre link to Sub-Saharan Africa (SAT-3/WASC/SAFE) which provides connections to Europe (via Portugal) and the Far East for eight countries (Senegal, Ivory Coast, Ghana, Benin, Nigeria, Cameroon, Gabon and Angola) along the West Coast of the Continent and south to the Cape in South Africa. A second segment, in the Indian Ocean, connects South Africa to Malaysia while passing through Mauritius and India (SAFE). Jointly funded by 36 members and spearheaded by South African Telkom which invested US$85 million for a 13 per cent stake, the project cost about US$650 million dollars. The cable was expected to lead to much reduced international bandwidth costs, but so far this has not occurred due to the business models used to develop the project. Even the few countries that have access to international fibre through SAT-3 are not seeing the benefits because it is operated as a consortium where connections are charged at monopoly prices by the state owned operators which still predominate in most of Africa, and in many other developing regions. Landlocked African operators who have tried to purchase international fibre capacity directly from one of the consortium's international members have found themselves being charged as much to reach the SAT-3 landing point as they were charged to get from the landing station to Portugal. Sadly, the high costs have made it cheaper to send the traffic directly by satellite, even for SAT-3 shareholders such as Telecom Namibia, which has no landing point of its own. Except for some onward links from South Africa to its neighbours, and from Sudan to Egypt and from Senegal to Mali, the remaining 33 African countries are unconnected to the global optical backbones, and depend on the much more limited and high-cost bandwidth from satellite links.
Mike Jensen
In fact prices have barely come down since it began operating in 2002 and are sold at satellite prices of $4-8K/Mbps/mo even though the capacity is only 5% used. As a result the lack of fibre and lack of competition on SAT-2, international bandwidth to African countries, as seen in Figure 12 lags well behind most of the rest of the world.
It should also be pointed out that the fact there is only one fibre optic cable means the only backup is satellite which may not be configured to take the re-0reouted traffic, and in any case may have inadequate capacity.
One result and immediate cause of this is the cost of this are costs of Internet connections in Africa and how they relate to income (Compare Figure 13a and 13b) and affordability. For example "Currently \[2007\] prices on SAT-3 are up to US$15 000 / Mbps/month, while it is estimated to cost the consortium only about US$300/Mbps/month". From a posting by Dewayne Hendricks on Dave Farber's Iper list \-\- Bill St Arnaud. |
Figure 13a: 2002 |
Figure 13b: |
|---|---|
|
|
Market Research conducted by Paul Budde Communications indicates that 1 year of Internet access is greater than the annual income of most Africans. Only in Egypt, Libya, and Mauritius is the annual cost of Internet access < 10% of the annual income. A study on Promoting African Research & Education Networking (PAREN), sponsored by IDRC, in Jan 2005 showed that African universities pay on average 50 times more for bandwidth than, for example, U.S. universities ($5.46/ Kbps/month vs. $0.12/Kbps/month). W. Africa pays on average $8/Kbps/month, and N Africa $0.52/Kbps/month. An OECD study on Broadband in November 2007 showed a median for OECD countries of $16/Mbps/month with Japan at $3.09/Mbps/month. Even a recent Bandwidth Initiative by a c alition of 11 African Universities (MZ, TZ, UG, GH, NG, KY) + four major US Foundations to provide satellite thru Intelsat though it reduced the costs by a factor of three they are still at $2.23K/Mbps/month. Thus Africans are paying hundreds of times what is being paid by residences in OECD countries.
In 2006 Africa had close to 1 billion people or about 14% of the world's population. Its usage is also growing faster than the rest of the world. However, the Internet penetration shown in the table below is only about 3.6% so it is a huge potential market.
At the same time there are promises of considerably increased fibre connectivity to sub-Saharan Africa in the next two years (in particular in time for the Soccer World Cup in South Africa in 2010) as can be seen in Figure 14.
Figure 14: |
|
|---|---|
|
|
Also the UN, governments such as China, the UK, Europe, the US and companies such as AMD, Intel,Microsoft, Cisco, Nokia and Ericsson are recognizing the opportunities and needs and investing. This will introduce challenges of new development models such as more inclusive business models; bottoms-up approach; working in new regulatory, policy and poor infrastructural availability environments; working with governments and others to ensure fibres are installed with any major relevant projects (railways, roads, electricity pylons etc.); micro-payments; content in many new local languages; use of wireless for last mile connections; Internet kiosks and cafes, etc.
The PingER project is arguably the most extensive active end-to-end Internet performance measurement project today. It data goes back over a decade and more recently has focused on measuring the Digital Divide. It covers measurements from over 16 countries to over 150 countries (see Figure 15) containing over 99% of the world's Internet connected population. 45 of the countries are in Africa.
Figure 15a: PingER Deployment Dec 2007, red are monitoring sites, blue beacons that are monitored from most monitoring sites and green the monitored only sites. |
Figure 15b: Google map of African sites Dec 07 |
|
|---|---|---|
|
|
Information on Calculation MOS can be found at http://www.slac.stanford.edu/comp/net/wan-mon/tutorial.html#mos
For African countries the PingER metrics are shown for African countries ordered by sub region (e.g. N. Africa) and country, measured from Jan-Sep 2007 in the figure below. It is also seen how the various components contribute to the derived throughput (8 * 1460 / (Average_RTT * sqrt(loss))
One way of showing how Africa is behind the rest of the world from the point of view of PingER Internet measures is to show maps of the performance for several metrics for January through September 2007. These are shown in the Figures below.
Red indicates geostationary satellite |
Voice & Video de-jitter required |
Host & Network Fragility |
Data Transfer |
|---|---|---|---|
|
|
|
|
The figure below shows the correlation between International bandwidth Inn Figure 12) and the normalized derived PingER throughput. It can be seen that the correlation is good (R^2 ~ 0.59). It is also seen that the African countries (in Magenta) lie towards the bottom left where both the throughput and International Bandwidth are low. Libya is an anomaly since the only site monitored there is the email server for the Libya Telecommunications which should have good connectivity compared to most sites in Libya.
To verify whether the order of magnitude of the normalized derived throughput is correct we compare it with the Speedtest results from www.zdnet.com.au/broadband/results.htm. This application sends a known amount of data between your computer and servers in Australia. Then it measures throughput and besides returning the result to the user, it also saves and makes available the results by country and ISP. The correlation is shown in the figure below. The orders of magnitude of the two methods agree and there is a strong correlation. As expected the African countries have the poorest results.
There are many indicators of countries' development from organizations such as the International Telecommunications Union, the United Nations, the CIA, the World Bank try to classify countries by their development. They are quite challenging and require careful consideration of: what can be measured, how useful is it, how well it is defined, how it changes with time, how it changes from country to country, the cost of measuring, the time it takes to gather (to avoid being too out of date, how subjective they are.
Typically they are based on some combination of GDP, life expectancy, literacy, education, phone lines, Internet penetration etc. Some examples include the Human Development Index (HDI), the Digital Opportunity Index (DOI), the Opportunity Index, the Technology Achievement Index etc. In general agree with one another (R^2~0.8). More information can be found here. For this report we focus on just two, the DOI (since it is recent and covers more countries than most) and the HDI since it different from the others and focuses on the human condition,
The Digital Opportunity Index (DOI) measures countries' ICT capabilities in infrastructure, access path and device, affordability and coverage, and quality. It covers 180 countries, and has recent (data 2005, announced in 2006). There are 11 indicators: that cover mobile telephony coverage, Internet tariffs, #computers, fixed line phones, mobile subscribers, Internet users all in relation to the population of the country. A map of the values for world countries and the correlation with the normalized derived PingER throughput is shown in the figures below. The Correlation is fairly good. As usual Africa is the worst off being 30 time worse than the US in terms of loss.
"The Human Development Index (HDI) is the measure of life expectancy, literacy, education, and standard of living for countries worldwide. It is a standard means of measuring well-being, especially child welfare. It is used to determine and indicate whether a country is a developed, developing, or underdeveloped country and also to measure the impact of economic policies on quality of life." from Wikipedia.
The figure below shows the correlation between the HDI and the normalized derived PingER throughput for Mediterranean and African countries. The correlation is seen to be strong (R^2~0.64). It is also seen that N. Africa has 10 times poorer performance than Europe, yet N. Africa is several times better than say E. Africa. E. Africa is poor, limited by its satellite access. For W. Africa there are big differences, some such as Senegal can afford SAT3 fibre while others use satellite, There is great diversity between and within regions. This is also show for the normalized derived PingER throughput for Africa in the figure to the right where one can see the overall throughput performance is poor to bad. There is a factor of 10 difference between Angola and Libya. N Africa is the best, and E Africa the worst. There are big differences within regions. In 2002, BW/capita ranged from 0.02 to over 40bps - a factor of over 1000
The Corruption Perception Index (CPI) relates to perceptions of the degree of corruption as seen by business people and country analysts, and ranges between 10 (highly clean) and 0 (highly corrupt). The correlation between the Normalized derived throughput and the CPI is shown below. The overall correlation of R^2~0.55 is good, however if one looks at the correlations for a region such as Latin America, South Asia or Africa it is very poor and negative in the case of Latin America.
According to Mike Jensen, in 2002 ?there is almost no intra-African Internet connectivity and the vast majority of international bandwidth lands in the G8 countries - principally North America followed by Europe (Belgium, France, Germany, Italy, Netherlands, Norway, Portugal, and the UK). High intra-regional telecom prices have limited the establishment of links between neighbouring countries to just 5 - Gambia-Senegal, and South Africa's links to Namibia, Lesotho, Swaziland and Botswana. As a result increasing amounts of intra-African traffic must be transited through high cost cross-continental links.?
Looking at the figure below of the routing from the best connected country in Sub Saharan Africa, i.e. South Africa to the rest of Africa in 2005 it is seen that aprt from to Zimbabwe and Botswana traffic was routed via other continents. Not only does this increase costs but it means that In essence the African community is subsidizing the international carriers.
What is needed is to put into place International eXchange Points (IXPs) between countries to reduce costs and improve performance. This requires users (universities, countries) to band together to leverage influence, get deals etc. There is evidence of this happening in Ubuntunetand the Bandwidth Initiative referred to above. However, current providers (cable and satellite) have a lot to loose and many of these have close links to regulators and governments (e.g. over 50% of ISPs in Africa are government controlled), so this will be difficult in many cases and may require government intervention. Also the regulatory regimes on the whole are closed and resistant to change, and
sometimes ISPs themselves are unwilling to co-operate
A survey by Les Cottrell and Monique Petitdidier at the IHY meeting Ethiopia in November 2007 of leading Universities in 17 countries showed the following:
Other examples from Heloise Emdon at the Acacia Southern Africa UNDP Global Meeting for ICT for Development, in Ottawa 10-13 July 2006 include:
Also from the Harvard African Higher Education Study:
In the industrial age a pre-requisite was an expensive infrastructure of roads, railways, ports to distribute raw materials and products, plus factories (e.g. steel making, cars etc.). In today?s information age: help, technical information is readily accessible online, so one person with a computer and an Internet connection can be productive (e.g. outsourcing of help, information services etc.) An Internet connection is a necessary pre-requisite and cheaper to install than a highway. To compete in today?s world, investment in IT and networking is critical to enable opportunity and productivity. It can also help minimize brain drain by making Africa more attractive to trained people.
It is clear that there are many Digital Divides, by region, country, within countries, age groups, city vs rural etc. How one tackles the DD varies from country to country, region to region etc. Typically we (International Committee on Future Accelerators (ICFA) / Standing Committee on Inter-regional Connectivity (SCIC)) have seen for Brazil, Romania, Pakistan and India it involves :
Maps of Connectivity from South Africa
An interesting way to analyze the state of internet connectivity in Africa is to look at colored maps of various metrics as seen from our monitoring node in South Africa. 
Ping Unreachability Packet Loss Min_rtt
Routing in Africa seen from SLAC
Using programs written to analyze traceroutes for different regions we have been able to generate topology maps of various regions in Africa as seen from SLAC. These maps provide a valuable insight into the various ASNs that are encountered as packets are routed from SLAC to Africa.
SLAC to West Africa SLAC to South Africa SLAC to East Africa
SLAC to Central Africa
GEANT has connections to EuMed in particular Marocco, Algeria, Tunisa and Egypt, see http://www.dante.net/upload/pdf/EUMED-poster.pdf . They are now working on connecting to Ubuntunet East and Ubuntunet South.
The SAT3 fibre serves some countries of Western and Southern Africa. The EASSY fibre is intended to assist Eastern Africa.
ICT in Africa: a Status Report, Mike Jensen
West Africa Submarine cable Connection
Web Atlas of Regional Integration in West Africa
African Scientific Network
African Universities
Columbia University Africa Studies
The bandwidth Initiative: Opening the power of the Internet to African Universities
University of Zambia Status.
