Tag Archives: broadband

How To Make The Digital Dividend Pay Out In Africa


Current television spectrum allocation

Digital Divide, Digital Dividend, Digital Yadi-yadah.  You would be forgiven if the term Digital Dividend didn’t immediately resonate with you given the proliferation of all things “Digital” in recent years.  A quick reminder then.  The Digital Dividend refers to the spectrum that is freed up in the conversion from analogue television broadcasting to digital broadcasting, also referred to as the Digital Switch-over (DSO), a change that has largely already taken place in the industrialised world and is slowly gathering pace in Africa.  This involves deploying digital transmitters to replace the analogue ones and either new digital televisions or digital Set Top Boxes (STBs) for existing televisions.  But the fact that more channels may become available or that you can receive television channels optionally in high-definition is less exciting to me than what can be done with the spectrum that is freed up. How much spectrum is being freed up?  There is over 400MHz of television spectrum.

Digital broadcasting uses a fraction of the spectrum that analogue broadcasting does and in Africa, there are few enough analogue terrestrial television channels per country to begin with.  What is more, it turns out that television spectrum exists within a very attractive range of the frequencies. What makes a frequency attractive?  Propagation or the ability of a radio wave to go through obstacles.  The lower down you go down the spectrum, the longer the radio waves and the less they are inclined to bounce off solid objects.   That means that you can cover a larger area with a single transmitter and that means that the cost of building a communication network drops significantly.  There are trade-offs however.  Longer waves carry less information so you can’t pack as much data into the same channel but that is a very reasonable trade-off when it comes to planning rural networks where the cost of network deployment may be a bigger issue than ensuring >20MB/s download speeds. So the Digital Dividend spectrum is extremely appealing from an infrastructure cost-of-ownership perspective.  It is unfortunate then that most of the debate around the Digital Dividend has largely been held within the broadcast community.  Not that digital broadcasting isn’t important but the Digital Dividend is an extremely valuable resource that needs to be considered holistically in terms of its national strategic value.

At the World Radio Congress (WRC-12) last year, there was confirmation of 790-862MHz (popularly known as the 800MHz band) as a global IMT (mobile) band. There was also a move by some African countries to have the 694-790MHz band (popularly known as the 700MHz band) made available in Region 1 (Africa and Europe) on an accelerated basis, probably because there are lots of CDMA players already in the 800MHz band. 700MHz is likely to be confirmed as an IMT band for Region 1 at the WRC in 2015.  So that’s good news for mobile operators except that the release of the 700MHz and 800MHz bands is being treated as contingent on the completion of the DSO.  Given the delays that have plagued the DSO on the continent, this seems like a dangerous strategy.  Why can’t one or both of these two new IMT bands be cleared for use while the DSO is going on in the lower end of the television spectrum?  At the very least, preparatory work for release of this spectrum ought to be going on now.


What a future allocation of spectrum might look like.

But the situation is worse than just a disconnect between the broadcaster and mobile operators.  There is also the prospect of missing out on an alternative access technology that could make a real difference for rural access.  Television White Spaces technology has the potential to create a vibrant rural access industry in Africa.

Television white spaces refers to the guard bands left between analogue television broadcast channels in order to prevent interference. TV White Spaces technology is capable of serendipitously re-using that empty spectrum without interfering with existing television broadcast. The initial vision was that through spectrum sensing, the devices would automatically use whatever empty spectrum was available, as a secondary user. That means if a television signal suddenly turn on in a frequency being used by a TV White Spaces device, it would automatically cease using that frequency and find another empty frequency to use. The broadcast and wireless microphone industry in the U.S. were not satisfied with this solution and the concept of an geo-located authentication database was introduced whereby TV White Spaces devices would need to authenticate against a spectrum database to see what spectrum was available for use in the area it was being used. Very low power TV White Spaces devices are still allowed to use just spectrum sensing. In general TV White Spaces regulation in the US has been the victim of massive lobbying and the result is some extremely hamstrung regulation.

The UK has largely followed the US regulation with one significant improvement. The power output level of the devices is not fixed but can be dictated by the settings in the authentication database. This means that higher power output levels could be assigned in sparsely populated rural areas versus areas where there are many other spectrum users.

What is exciting about this technology?

  1. No spectrum license required or at least a very nominal one. This means new opportunities for small entrepreneurs to provide alternative access.
  2. Great propagation. A typical TV White Spaces link can go 8-10km without any effort and is not obstructed by trees, buildings, etc.
  3. Innovation. WiFi has gone from a niche spectrum for experiments to an industry that is expected to be worth over 6 billion dollars in 2015. ~70% of smartphone data traffic in the rich world goes over WiFi. This is what open spectrum offers. TV White Spaces has the potential to be another such industry because of the low barrier to entry.
  4. No spectrum re-farming required. Because TV White Spaces technology is designed for secondary use of spectrum, there is no need to move the primary spectrum holder. This is a quick and easy win. Conflicts can be easily resolved by the regulator thanks to the authentication database.

TV White Spaces are finally gaining traction however.  Google is sponsoring a pilot in South Africa and in Kenya, Microsoft are supporting a pilot in partnership with the Kenyan government and a satellite operator there.  Those are good signs but in general the discussion of the Digital Dividend has been trapped in bureaucratic silos.  There needs to be a broader acknowledgement of the strategic value of the Digital Dividend and a strategy that addresses it holistically.

AfTerFibre Update October 2011

This is a short summary of progress and learning from the first couple of months since launching AfTerFibre, a project to map terrestrial fibre projects in Africa. From the beginning AfTerFibre has been designed as an open project both from the point of view of transparency and from the point of view of participation. So the first goal was to make it easy to share information. This involved setting up a AfTerFibre Google Group so that anyone could contribute information or ask questions.  Next we needed a place to store resources as we found them so a wikipedia page was set up to capture information and links to maps of terrestrial fibre projects.  That part has gone reasonably well.  We now have 83 people in the Google Group and the Wikipedia page now has 67 African  operators known to have fibre projects listed, for which about half have links to maps.  Where possible I have linked directly to the map on the web if it exists.  In other cases, I have uploaded map images that I have found to a Flickr set.  Finally there is also a Diigo list of news links related to AfTerFibre.

Having gotten the repositories for the raw information in place, the next challenge was to find out how to create an information chain that would make it easy not only for people to contribute map information but also to submit updates. I’ve been working with some of Google Africa’s GIS team in Nairobi to solve this.

Step 1 – Convert Image Maps into GIS Maps

This stage of the mapping turns out to be remarkably simple. GoogleEarth (GE) is an amazingly simple yet powerful tool. You can import any jpg or png image of a map as an overlay into GE. GE makes it a piece of cake to stretch the image to neatly match the image map borders with the real GIS borders in GE. This makes it relatively simple to then trace fibre routes with the Path tool. I really can’t overemphasise what an amazing job the Google Earth designers have done to make it easy to trace maps. As an example, when you are actively tracing a path in GE, the mouse is no longer available to move the geography as you trace along a line. Happily the keyboard arrow keys are enabled for this purpose allowing you to trace with the right hand and move the geography with your left hand. More software should be this well designed. Because I am a GIS greenhorn, I have just assumed that KML is the appropriate format for this data.

Step 2 – Create a online updateable repository of map data that can feed the AfTerFibre Map

Our initial assumption was that Google Fusion Tables would make a good data repository. It is capable of directly importing the KML files generated from the tracing of maps in Google Earth. The map you see above is a direct representation of the AfTerFibre maps after a simple import into Fusion Tables.

Here is where things get more complicated. Fusion Tables turns out to have some idiosyncrasies (it’s still in Beta) which make it somewhat problematic for AfTerFibre. For a start, it doesn’t allow for easy multi-row editing so if you wanted to make a change across an entire country or operator, it is a very slow process. Second, while Fusion Tables allows you add fields, which is essential for AfTerFibre to capture all the additional information on fibre projects, Fusion tables stops importing KML files properly after you change the database structure. The geo-coordinates are mapping into text fields etc. Chaos and confusion.

An alternative then is to import the KML files into Fusion Tables and directly export as comma separated files (CSV) files which can then be imported into a spreadsheet program. A spreadsheet solves some of the problem by allowing multi-record editing and also making the data easier to massage in general. Each time a new fibre map is made, it could be imported into Fusion Tables, saved as CSV, imported into a separate spreadsheet and edited to match the field layout of the master spreadsheet and the imported into the master spreadsheet. The master spreadsheet can then be exported to CSV which can then be re-imported into Fusion Tables for display in Google Maps.

If this were a one-time process, the above might be an acceptable solution but not only will new maps regularly be added but existed maps are likely to receive numerous corrections.

So, what to do? It seems to me that what is needed is some kind of versioning repository (like Github) for the KML files for AfTerFibre. I’ve done some googling around this but haven’t come across anyone who is storing their KML in a versioning repository. It seems like a clever idea but it worries me that I can’t find anyone else doing it.

If KML files could simply be updated in a Github or similar repository, then it wouldn’t be that hard to write a script that retrieved the files from the repository and massged into a GIS display with the appropriate mouseover information displays etc.

So, I’m a little bit stuck here and open to suggestions.

Vodacom – Oh Really?

Vodacom Group CEO Pieter Uys seems like a nice guy.  I’ve never met him but you look at his picture and you think, here is a decent fellow.  Which makes it all the harder to credit his defence of Vodacom’s pricing.

He spoke recently at the launch of a World Wide Web Foundation and Vodafone sponsored report entitled ‘Making Broadband Accessible for All‘.  When questioned by ICASA General Manager of markets and competition, Pieter Grootes, about the significant disparity in pricing between Kenya and South Africa, a question hopefully inspired by this post, he had this to say:

“It’s not possible to have a call that lasts long in Kenya. And it’s not possible to have fast mobile broadband speeds at the same rate as SA. There’s a direct relationship between pricing and quality.”

He appears to be arguing that Vodacom has 9 times the call quality that Safaricom does and can thus justify charging 9 times what they charge.  Now call quality in Kenya may not be perfect but I think it would be very hard to come up with real evidence to support an order of magnitude quality difference between South Africa and Kenya.

More to the point, why does Vodacom get to choose what kind of call quality we have?  Why isn’t this something that the market establishes?  Why can’t people choose a cheaper, lower quality network or a more expensive, higher quality network? The answer is simple.  In South Africa, there is no market.  There is just an uneasy alliance of rent-seekers.  Just like in the world of professional wrestling there is a lot of bluster and talk of competition but when it comes to actually stepping up and squaring off with other mobile operators, the outcome is predetermined, the game fixed.

There are signs of hope though.  8ta’s recent dramatic drop in broadband charges is a possible sign of good things to come but the fact that it is only broadband that has been affected and that it is a temporary offer does not really represent the kind of tectonic shift in pricing needed, but you never know.  It might be the gust of wind that presages a storm.

Interestingly, at the same session, Uys found an ally in researcher Winfred Mfuh, who said that

“constructing and renting a base station can be three-times the cost in SA than that of Kenya, where there are a lot of price considerations.”

I accept that putting up infrastructure may be cheaper in Kenya than South Africa, however, I doubt it needs to be three times more expensive in South Africa.  Mobile operators in India have mastered the art of outsourcing infrastructure.  Vodacom’s infrastructure is expensive because they’ve never had real market pressure on them to optimise their costs.

This is not idle speculation.  South Africa’s expensive, uncompetitive telecommunications environment is well documented by researchers.  Something worth considering closely when considering how new spectrum should be auctioned.  A set-aside for new market entrants is a minimum consideration.  In Canada, incumbent operators fought tooth and nail to prevent Egyptian-owned Wind Mobile from getting access to spectrum in the AWS auction two years ago.  I see Wind Mobile now have the best mobile broadband prices in Canada.  What a surprise.

Open Access, Africa, and Yochai Benkler

Since the announcement of the EASSy undersea cable in 2005, Open Access has been a term of significant debate in the development of undersea cable initiatives and in the general strategic development of communication infrastructure in Africa.  Yet, it is a term that is variously understood and often abused especially by the marketing departments of undersea cable initiatives.  There is not an African undersea cable initiative that doesn’t claim to be Open Access but all operate on different ownership and pricing models.

Put simply, Open Access embodies the principle of enabling competition at every layer and every point of the communications infrastructure.  Communications infrastructure depends on a chain of access and if even one link in the chain is controlled by a single entity, then the entire network is vulnerable to monopolistic rent-seeking behaviour.  Thus to have Open Access, there must be competition in the provision international fibre access, national backhaul solutions, and last mile services, whether fixed, mobile, or nomadic.  Not to mention competition in the provision of services over that network whether content or transaction related.  The opposite of Open Access are vertical markets controlled by single entity or a small cartel of entities.  A good example of this was as recently as 2007 when Telkom South Africa controlled access to the then only undersea cable, SAT3, and also controlled the fixed-line local loop.  Now we have two and soon three undersea cables and a form of local-loop unbundling for Internet service provision.

On Wednesday morning, I shall have the good fortune to engage Yochai Benkler in a discussion on the topic of Open Access infrastructure in Africa and inspired by the range and thoughtfulness of the feedback that Erik Hersmann obtained recently when he asked “What would you say to Nokia Africa?“, I thought this might be another opportunity crowdsource some perspectives on Open Access in Africa.  For the record, I am not referring to the Open Access initiative for academic publishing.  That is an excellent initiative but unrelated to Open Access infrastructure policies.

Yochai Benkler, author of the Wealth of Networks, was also the Principal Investigator for a comparative study of broadband initiatives carried out by the Berkman Center and commissioned by the U.S. Federal Communications Commission.  Entitled “Next Generation Connectivity: A review of broadband Internet transitions and policy from around the world“, it reviewed efforts, mostly in OECD countries, to strategically develop pervasive, affordable, high-speed access infrastructure.  A key finding of the study was that countries that have successfully implemented Open Access policies tended to have better and more affordable access.

That brings to Wednesday.  Yochai’s experience is more in the rich world.  My goal is to provide an overview of attempts, successes, failures to implement Open Access policies in Africa.  I know some of these stories but it would make the conversation that much more interesting to have other stories to draw on.  In exchange, I promise to post back what I’ve learned from our conversation.

Help me out, please?

Municipal Broadband Investment in South Africa

There’s an interesting AFP article on the city of Johannesburg’s 1 billion Rand (~105 million USD) contract with Ericsson to deliver municipal broadband infrastructure.  In it they quote an anonymous “telecoms expert” who says:

…that municipalities were already stretched to provide basic utility services, and that the provisioning of telecoms services would simply put additional strain on already ineffective organisations.

He further argued that government’s involvement in telecoms in the past, typically through Telkom and Sentech, had hurt the industry and kept South Africa back, and that there was no reason to believe that things would be different this time.  He suggested infrastructure investment incentives such as tax breaks to operators rather than direct government involvement in such networks.”

Given that the city of Johannesburg is giving a billion Rand to Ericsson, I don’t think the straining already ineffective organisations argument is really what’s at issue.  I suspect straining already over-stretched municipal budgets is probably more to the point.  However, even that is secondary to the two most important weaknesses in Johannesburg’s approach:

1) Johannesburg now has its hand in the cash register

By betting all or nothing on Ericsson, the city has aligned its broadband fate with Ericsson.  The city obviously wants to see Ericsson succeed, which is all well and good. However, the dark and ugly flip side of this is that the city will not want competition in the municipal broadband market to flourish as it might hamper the success of their investment with Ericsson.  Sound familiar?  ::cough::  Telkom, 1990s, hello?

2) Monolithic tenders are often ridiculously inefficient

Around the world we have seen the failure of municipal wireless networks in the last few years.  The nature of these failures is a matter of wide speculation but I would argue that a major cause is large cities betting their budgets on a single provider that promises the world.  Earthlink’s failure in the United States is a prime example of this.  One cannot help but wonder how much of that 1 billion Rand is going to be eaten up by endless consultant fees, meetings, overheads, etc.  How much will actually go to deploying infrastructure?  If Ericsson is going to deliver affordable cost-effective broadband for the city of Johannesburg, they are going to need to be as lean as the leanest Seacom-connected ISP start-up.  This strikes me as unlikely.  Equally, broadband is a very fast moving technological field. What happens when all of Ericsson’s equipment is outmoded by leaner, smaller competitors?  Will they come back to the city with their corporate hands outstretched for another billion Rand to upgrade?

Hearts in the Right Place

Having said that the hearts of municipal officials are clearly in the right place.  Johannesburg councillor Parks Tau says, “We regard access to broadband as a key driver of economic growth and wealth generation.”  Jacquie Subban, head of strategy for the eThekwini Municipality, argues the importance of broadband access in terms of economic growth and social development and says that broadband infrastructure should be made available cheaply.  They are SO right.  Well and affordably connected cities can be significant drivers of social and economic growth.   The key issue is where does this correct and well-intentioned thinking ultimately lead municipal governments.  Failing to learn the lessons on offer from the likes of Sentech, Telkom, Infraco, is frankly hard to excuse.

So what should municipalities do?

  • Focus On Enabling Infrastructure. The OSI model is a great guide for municipalities.  Anything above Layer 1 should be questioned.  Municipalities should make a point of investing in the most basic infrastructure possible that lowers barriers to entry for industry and enables competition as opposed to competing with industry. Cities are in a unique position to assist in the deployment of fibre infrastructure.  They already dig up the streets on a regular basis. No road should be re-surfaced, no sewer pipe replaced, no traffic light installed without consider options for the deployment of at least conduit and possibly fibre.  When a city deploys fibre, outside of servicing their own needs, they should let others light their own fibre.  Access to dark fibre should be non-discriminatory.  The City of Stockholm is a great example of how a city can partner with the private sector to deliver pervasive affordable broadband.  The City of Cape Town appears to get this.
  • Create Incentives for Sharing. Where possible either create incentives or mandate the sharing of ducts, in-house wiring, etc.  Encourage cost-sharing for new infrastructure deployment.  Promote transparency by making all existing and planned deployments a matter of public knowledge.
  • Be Proactive. One obvious idea that comes to mind would be for a city to carry out its own spectrum audit.  Find out what spectrum is in use and lobby for local access to unused spectrum. A recent survey of spectrum use in Dublin, Ireland revealed that average use across the primary spectrum bands was less than 14%.
  • Be Creative. Street lights, tall buildings, existing urban features are all possibilites for deploying infrastructure.  The way that the city of Paris used the extensive sewer system to deploy fibre is a brilliant example of broadband innovation.

I believe that the successful model for pervasive, low-cost municipal broadband involves a mix of large infrastructure providers and entrepreneurial smaller ones to keep them on their toes.  The above list is just a start.