Research

An invasion of armies can be resisted, but not an idea whose time has come
Victor Hugo, Histoire d’un crime, 1852

It's always good to know why you are doing something. I'm a firm believer in PhD's that are strongly grounded in real-life problems that are facing the human race. So before leaping into the actual wireless mesh networking research that I am focusing on, I am going to go on a philosophical journey and try and build a motivation for my specific focus on rural wireless mesh networks.

For the sake of the argument, substitute rural wireless mesh networks with bringing the Internet to rural areas, although there are benefits other than simply Internet access, just Internet access itself will build a strong enough case. There is a lot to reflect on when thinking about why bringing Internet connectivity to deep rural areas in places like Africa is a good idea. There are many people who would argue that there is no point giving people Internet connectivity when people are dying of malnutrition, TB and Aids. Surely running water, health care, shelter and food security should be the key focus before something as advanced as broadband access is even considered. This is not an easy question, the "one laptop per child" project asks the same question "Why give a laptop to a young child who may have no electricity or running water?" Their answer is simply, "If the laptop is their portal to an education, you don't stop education until all other challenges like food shelter and water are solved, you solve them at the same time because education is the foundation of the solutions to the other problems"

One can almost hear the echo of the elite literate class in the middle ages asking why rural peasants would possibly ever want to read books or be allowed to vote on the countries future. There are many parallels between societies in developing countries moving from an Agrarian to a knowledge based economy using Information Technology and the development of widespread literacy and the enormous social and intellectual changes of the Enlightenment period due to the invention of the printing press.

Well, if every child in the developing world now had a laptop would that be good enough, do we still need to connect them to the Internet? Perhaps we need to step back a little to reflect on what has happened in the past to help understand the future.

Progress and innovation in any civilization has always hinged on the free flow of information between people and true freedom is achieved when no voices are silenced by governments or powerful individuals. Consider countries such as Burma, North Korea, China and recent events in Zimbabwe, the first target of any dictator is preventing the free flow of information and silencing the media. This leaves the population subject to massive exploitation and trapped in a poverty cycle. As the age of the printed book and the free flow of knowledge between all people, not just a few elite, began with the Gutenberg movable type system in the 1400's, so the age of the Internet in the 1990's has unleashed a new era of information exchange between mankind like nothing seen before. It is becoming possible for all people on the globe to have equal access to the same knowledge resources as more books are digitized and more knowledge is captured. Of course that assumes that all people have equal access to the Internet which is clearly not the case when you look at a graph like this:

ICT Levels around the world, by region in 2007 (http://www.itu.int/ITU-D/ict/statistics/ict/index.html)

It is clear that users living in the developed world have penetration rates which are far in excess of those in developing nations especially when it comes to Internet and broadband access. But Africa is also trailing penetration rates of other developing nations in Asia. This is the digital chasm that could prevent Africa and other developing nations from being part of the global knowledge economy unless drastic measures are taken.

From another angle, when one thinks about inventiveness and progress that result from free information exchange, Lawrence Lessig has this to say about the importance of free resources for innovation in his book, Future of ideas. "Always and everywhere, free resources have been crucial to innovation and creativity; and without them, creativity is crippled. Thus and especially in the digital age, the central question becomes not whether government or the market should control a resource, but whether a resource should be controlled at all. Just because control is possible, it doesn’t follow that it is justified."

With this as a backdrop, it is clear that people living in rural areas are being left behind. Is it such a bad thing to be left behind? Shouldn't people who are quite content living off the land on a farm just simply be left to enjoy their simple life on their farm. Well, what happens when a farmer has repeated crop failure of his maize crops with the current soil and rain conditions, or they need to go to a specialist hospital in a far away town to diagnose a strange skin rash or there are no mathematics or science teachers available at the small rural school. Could connectivity to the Internet help with some of these problems?

A quick search on an agricultural web site mentions that the soil conditions and rainfall patterns in that area would be ideal for sunflower growing (true story here), a nurse at a local clinic emails a digital image of the skin ailment to a specialist at a city hospital receiving instructions on what topical treatment to administer (true story here) , a mathematics teacher at a city school offers to web cast her lesson to the rural school everyday and can interact with students in the school via video conferencing facilities (true story here). It is possible to modernize without having to urbanize to improve peoples quality of life and access to knowledge. Knowledge brings opportunities, which means that instead of someone being trapped in a cycle of poverty, they will have the resources and skills to transform their own life and the community around them.

From a political point of view, mesh networks create a system which links people together using a decentralized structure which is hard to control. It could enable people in countries where the telecommunications is being controlled to build under-ground networks where authorities have no control over content that is communicated over the network. It isn't owned by one single entity and the entire community owns the infrastructure which makes it hard to shut down. If some nodes are shut down, the network will automatically repair itself and find other routes between a source and sink.

Here are five key reasons why I think mesh networks are a perfect fit to brining connectivity to rural areas:

  1. I have seen it working in a number of projects I have been involved in - The Peebles valley mesh in Mpumalanga, Scarborough mesh in Cape Town and Linknet in Zambia.
  2. From a political point of view it puts control of the network back in the hands of the people, the community owns the network which makes it hard to shut down and prevents exploitation by government or dominant individuals or institutions.
  3. It is very robust to failure because it automatically repairs itself when a router in the network goes down.
  4. It is very cheap and getting cheaper - it runs on wireless routers which you can get for about $40 you can even boost the signal with cheap antennas made from tin cans.
  5. Router software and server software to manage the mesh is all available as open source software, so start up costs for someone who wants to run this as a business are very small.

Research areas

Mesh routing protocol benchmarking and development to support energy constrained devices and multi-media

I have done a large amount of work trying to understand the key differences between various open-source mesh routing protocols available (See masters thesis and paper on AODV vs DYMO vs OLSR and paper on Batman vs OLSR). But what I have realized is that protocols which become dominant depend little on rigorous scientific comparisons. This is mostly based on strong representation at IETF meetings as well as association with dominant academic or cooperate institutions. When new protocols surface there is the usual old protecting itself against the new. This has been very evident when looking at reactions to the recent B.A.T.M.A.N mesh routing protocol which was developed by a hacker community in Germany. It is clear that the debate is far from over even though 802.11s has standardized on derivatives of OLSR and AODV. I hope to one day do a final comparison of the mesh routing protocols in wide spread use: 802.11s, Srcr being used in the Meraki routers, OLSR v2, B.A.T.M.A.N., non-hop-count AODV such as AODV-ST and possibly some others on the 49 node wireless mesh grid at the Meraka Institute in South Africa. I will invite authors of the code to provide me with the most up to date source code as well as configurations they believe will be most optimal for the setup.

When mesh routers are running on renewable energy sources, routing paths need to be chosen to minimize the use of routers which are energy constrained. For example if one router is running on solar energy and another on wind energy then at night routes should be chosen which use wind powered nodes and in the day routes which use solar powered nodes. The routing decisions also need to be made aware of battery levels at nodes which are powered by batteries in order to choose routes which maximize the life span of the network.

Once TDMA-like MAC protocols are being utilized in a mesh network, it becomes possible to choose routes which are not only based on least congested paths but also those which have more slots available for real-time media flows. There is increasing interest in building village-telcos which make use of mesh to handle VoIP calls within a village and from a village to the rest of the world and supporting more VoIP calls will become increasingly important.

Auto-configuration to de-skill mesh networks

Traditionally telecommunication networks are highly engineered with careful planning to choose sites where radio towers are installed, what antennas to use, what angle to orientate and tilt the antennas, what power levels to use and how to engineer the IP addresses in the network. In a mesh network there is very little control over where nodes are placed and the network needs to autoconfigure itself as close to the optimal as possible. This include issues such as:

  • Optimal auto channel allocation on mesh routers that have multiple radios (Some existing solutions)
  • Optimal channel selection which minimizes the average amount of interference across the network on free open channels (Some existing solutions)
  • Auto IP address and subnet allocation schemes which avoid address collisions (this is mostly solved with IPv6 but other schemes are possible)
  • Gateway discovery mechanisms which optimally balance traffic in the mesh between multiple gateways being supplied by a single operator. Or gateway discovery which allows users to quickly discover and choose from multiple gateway providers selling services into a mesh (Not enough done in this area yet).
  • Optimal power level allocation to radios in the mesh is dependent on minimizing the amount of interference but at the same time maintaining a good level of redundancy in the mesh (some existing solutions).
  • Once the cost of using smart antennas is low enough, these can be employed to improve the spectral efficiency in a mesh and avoid unwanted interference. Algorithms to control elements of the smart antennas will need to be developed as part of the mesh routing protocols for direction of arrival and direction of departure of packets in the mesh (Very new area).