Sámi Network Connectivity A proposition to establish connectivity in communications

Sámi Network Connectivity A proposition to establish connectivity in communications challenged areas Avri Doria with materials by Maria Udén & Anders Lindgren, Luleå University of Technology and Ingunn Limstrand, Kvinneuniversitet Nord Photographer Lena Kuoljok-Lind The SNC System Combines Opportunistic Routing, Delay Tolerant and Ad Hoc Networking in order to Provide Internet Connectivity SNC History • Initiated among a group of Laponia reindeer herders • Most herders are men, though the group that initiated the project are women. • These women wanted to improve women’s possibilities for remaining within the Sámi society. • Their means was a gender equality project ”Kvinna i sameby” or KIS – in English ”Woman in the Sámi Village”. SNC History (cont’d) • KIS project workers contacted the Division for Gender and Technology at Luleå University of Technology. • An associated project was started by Maria Udén to help analyze social patterns and key areas for technical development. • Avri Doria, guest researcher at LTU at the time, heard about the cooperation and after consultations and together with Maria Udén, launched the SNC proposition during spring 2002. • Received first funding Jan 2003 from Vinnova – Swedish funding agency that promotes joint Governement/Indsutry funding of research programs SNC Today • Research project 2004-2006 funded by Vinnova • Gender equality monitoring project funded by EU Objective 1 • Network of women in Norrbotten who are active parts in ICT development (users and/or experts) – Founded a NGO as a spin off Nomadic Women's ICT Network • Business opportunities being investigated by locals • In final year of project – Good progress in meeting academic goals – Completed Lab integration and Testing – Completed basic field pilot – Working on wrap up documentation Top Level Project Objectives To Create Technology that can be widely used for communications challenged areas Create entrepreneurial opportunities for indigenous people in ICT To Support Internet Services Community access Business services Targeted Internet Services Community access Email Web access Educational opportunity Business service Herd telemetry Small business opportunities Web sales Software development and support Participation & Education are key goals • In order to make the system sustainable, • In order to even get the production system off the ground • Ultimately, it must end up a local project in order to be sustainable. Basic Constraints • • • • • • Semi nomadic population No wired infrastructure Inconsistent wireless infrastructure Expensive and intermittent satellite coverage Population wants/needs connectivity Very limited funding – Reindeer herding is subsistence farming – Since population is only semi nomadic and lives in the towns for part of the year, the local governments consider them as having network access Solution • Involves use of a mix of fixed and mobile nodes • Involves use of… – Delay Tolerant Networking (DTN) borrowed from NASA JPL – Mobile Networks – Mobile Ad Hoc Networking – 802.11 – Distributed web caching – Any existing technology that can be put into service • Involves making a change to the way a sub-network is connected to other networks and to the Internet at large SNC Technology Makes use of how the vast area is populated: not evenly distributed but dynamically SNC technology • The technical proposition is rather simple in its most basic terms • Yet its deployment is not be quite so simple. • It relies on the opportunistic idea that where communications infrastructure exists it should be used. • And where it doesn’t, or can’t, exist then mobile relays can be used. • These mobile relays can be mounted in the vehicles the people drive, helicopters, drones and can be carried in backpacks. • The communities send out email and requests for web content and have that content delivered. • In essence just as the postman delivers letters and magazines to a community, if they could deliver in these remote areas, the mobile relays bring content to the remote communities. More about SNC • To move from architecture to deployed system touches on many technical areas. • Some of these still require research while others require innovative application of existing technology. – The project set up • Ph.D. Level research projects • MA level engineering projects – Several Universities involved. All development being done in the context of student projects. • The core is prioritizing robust connectivity, not real-time connectivity. • Make use of architecture developed in DTNRG and uses bundle forwarding. Connectivity Based on Human Mobility • In SNC, connectivity is designed to rely on use of mobile relays along human pathways. • The probability that this will function in the vast grazing area that reindeer herders move within rests on the observation that concentrations of population occur throughout the year, at different places - clustering. • This population characteristic can be described as “global sparsity but local density”. • Object of a typographical study Human Mobility, (cont) • Much of the reindeer grazing lands are populated by 0.0 persons per square kilometer. • This indicates one would have to wait very long to meet some person or vehicle carrying a relay to further ones communication. • Yet, the statistical mean value does not describe how the area is in fact populated: not evenly distributed but dynamically. • The fact that the SNC connectivity solution, designed for one of the most sparsely populated regions of Europe, is socially based indicates how finding communication systems for challenged areas may succeed by thoroughly investigating presumptions about circumstances that are marginal in relation to dominating ways of life. • Results may have extensive other applications as diverse ICT for Peace, IPN, or ecotourism. Flexibility • The opportunistic routing which is the goal of SNC matches what is often required from people who live in sparsely populated areas: flexible, non-specialized, taking all possibilities into account. • Thus, the connectivity proposition harmonizes with strategies already practiced to meet the type of conditions for social and cultural reproduction it will serve. • This mode is also very nomadic; not building massive structures at every instance, but adding something small which is portable, to enable use of structures that are already established. Normal Internet Connection • In a normal Internet connection, the sub-network, in this case belonging to a community, is connected for the entire time that communication is ongoing. • For example: In an email dialog (simplified): – – – – – A person’s system (the mail client) sends an hello The mail server responds with its own hello The person’s system sends the message The mail server acknowledges receiving the message The message is delivered • In other words, the dialog between the client and the server takes place by taking turns across the network. This entire dialog must be completed within seconds. In Delay Tolerant Network • In a DTN the delay, because of a lack of connection, makes using the normal procedure impossible. – Delay can be minutes, hours or longer • In this case the mail dialog would be as follows: – Person’s computer sends Hello and the message – Some time later mail server accepts hello and message and sends acknowledgment – Some time later person’s computer receives acknowledgment – Could be minutes or even hours between each message – Within current protocol constraints, this does not work. In other words • Instead of having a normal network conversation – similar to a telephone call • Participant in a DTN uses a form of communication that is more like sending a letter. All the data and metadata is placed into a single Bundle and this bundle is forwarded through the DTN region. Or Back to early days of networking, e.g. uucp, but with greater functionality requirements Services to be offered in SNC • Program's focus – Support for basic email – Support for web cache services • Provider Pushed • User Pulled – Scheduled – Event Driven • Ad Hoc Request – Experiment with applications running over DHARMA in the DTN – A PlanetLab based mobility platform http://dharma.cis.upenn.edu/ SNC routing Based on Epidemic Routing • • Models the dissemination of data as the spread of a disease throughout the network. – As two nodes meet, they exchange information about the messages they carry. – Messages not previously seen are requested from the encountered node. – Floods the message throughout the network (subject to available bandwidth • High delivery probability/low latency • High resource usage • Best Path Routing is always achieved However • If buffer space and bandwidth were infinite, – Then Epidemic Routing would be optimal with regard to delivery rate and delay. • In current reality, infinite buffers and bandwidth are hard to find and indefinitely large ones are expensive. • Therefore, we wanted to do something smarter and looked for ways to prune the epidemic distribution tree. • One experiment involves the ProPHET algorithm. – A lot of simulation work was done showing the effectiveness of the algorithm – not discussed in this presentation. But there are papers. DTN Routing Method • Anders Lindgren et al. at LTU developed PRoPHET – Probabilistic Routing Protocol using History of Encounters and Transitivity for Intermittently Connected Networks • draft-lindgren-dtnrg-prophet-02 – Loosely based on Epidemic Routing • In Prophet - epidemic tree is pruned based on probability of delivery and messages are only passed to more likely nodes Example of Transitive Communications D •Source has bundles •Mobile node moves toward S S Time t1 Example of Transitive Communications D Mobile Node Approaches S Time t2 Example of Transitive Communications D Exchange S Time t3 Example of Transitive Communications D S Time t4 Example of Transitive Communications D S Time t5 Example of Transitive Communications D S Time t6 Example of Transitive Communications D S Time t7 Example of Transitive Communications D Message is passed to destination S Time t8 Example of Transitive Communications D Destination sends ACK S Time t9 Example of Transitive Communications D S Time t10 Example of Transitive Communications D ACK is propagated S Time tn Example of Transitive Communications D Timeout reached S DTN Routing Algorithm • For each node 0 <= P(a,b) <= 1; Initially P(a,b) = Pinit • Upon encounter P(a,b) = P(a,b) old + (1- P(a,b)old) * Pencounter • If no encounter within time threshold, decrease P P(a,b)= P(a,b)old * γk 0<= γ >= 1 is a an aging constant k - number of timeout units DTN Routing Algorithm (cont’d) • Transitive property If node A often encounters node B and node B often encounters node C then node C is a useful for forwarding toward node A P(a,c)= P(a,c) old + (1- P(a,c)old) * P(a,b) * P(b,c) * β 0 <= β <= 1 Scaling Constant for importance of transitivity • Note: β, γ, Pinit, Pencounter vary according to network conditions and are still object of simulation and experimentation. They will be different for different networks. Trying to discover guidelines. Forwarding Strategies • The time available for exchanging bundles may be limited, therefore testing various models: – Forward when there is a better chance of delivery • GRTR if P(B,D) > P(A,D) – GRTRSort in descending order of P(B,D) - P(A,D) – GRTRMax in descending order of P(B,D) – GRTR+ P(B,D) > P(A,D)&& P(B,D) > Pmax – Limit forwarding to a maximum number of times • GTMX if P(B,D) > P(A,D) && NF < NFmax – GTMX+ if P(B,D) > P(A,D)&& NF < NFmax && P(B,D) > Pmax Queuing Strategies • Since resources are finite the system needs to drop bundles. Some models for dropping bundles: – FIFO – MOFO – Most forwarded – MOPR - Most Probably Successful already – SHLI – Shortest time before timeout left – LEPR – Least Likely to be delivered successfully Note: there is a large topic of bundle custody and its effect of forwarding, dropping and queuing that I have not discussed here at all. Status of Protocol • Lego MindStorms implementation • Very successful demo at MobiCom 2005 – http://www.sm.luth.se/~dugdale/mobicom2005_final.avi – CDs with video of the demo are available for interested people (Give me a card and I will arranged to send one) • Draft compliant implementation in OmNet++ simulator. • Linux implementation in progress – Intended as a reference implementation – Implementation was tested in field pilot. – Currently being upgraded based on expereinced in the pilot test. • Keeping up with new architecture work being done DTNRG Example of a pending problem • Improved operation in partly connected networks and at high connectivity spots • We refer to this as: The Parking Lot problem. – What happens when multiple DTN relays are parked with new contacts coming and going throughout a time period longer then the timing intervals. – How does one differentiate between valid updates and invalid ones so as not to cause deterioration of the probability of encounter values • We have had lots of extended discussions about this. – Even took a while to convince everyone on the team that there was a problem • Different options are being explored. • But it is still an open problem Preliminary conclusions Some things we think can be learned from the SNC process About innovation… • Social and cultural knowledge/studies are key factors • Prejudices are less likely to lead to genuine innovation • People in Laponia are not evenly distributed with 0.1 person/km2 • ”Global scarcity but local density” About new actors and new partnerships… • The genuine novelty of the SNC partnership is one of its strengths • Yet, systems for funding tend to support the expected • Men still tend to decide about women’s future • Widely spread view say that Indigenous peoples and cutting edge technology are incompatible About women’s potentials… • Persons who are less prestigeous regarding technology can be more open to new possiblities • A person who is not included in established networks must be innovative • Less legacy from Industrial society brings solutions that are in pace with time http://www.snc.sapmi.net/