Recipient Institution:
Indian Institute of Technology, Bombay

KReSIT, IIT Campus, Powai,
Mumbai-400 076
India

Tel/Fax: 91-22-2576-7900

Project Leader:
Prof. Krithi Ramamritham
krithi@iitb.ac.in

URL:
http://www.iitb.ac.in/

Amount and Duration: US$ 25,579 / 16 months

Commencement Date:
1 October 2003

Location
Our research will be undertaken in the campus of Indian Institute of Technology, Bombay.

The pilot site is situated in the Pune district of Maharashtra state of India. Pabal is 20 kilometers away from the closest town (Rajgurunagar) and 70 kilometers away from the closest city (Pune). This site was chosen because of Vigyan Ashram - a school that caters for school dropouts has been doing some cutting-edge work in ICTs for development.

Duration
16 months

Project Background and Justification

Why ICT research in rural areas?

We believe that countries like India must lead the way for new IT markets, and indeed attempt to gain such leadership if it is to have a sustainable presence in the IT and Communications field. Our field research suggests that there is a huge untapped demand for computing in rural environments. However current solutions are not tailored for these markets, so the hardware and software industry will have to innovate and come up with appropriate solutions. Some of IIT-Bombay's current research projects, such as Media Lab Asia, have a strong focus on the needs of people who have not yet benefited from Internet and Communication Technologies. However none of them have a direct impact on networking infrastructure - the focus of this proposal. Hence the proposed work neatly complements the ongoing activities.
We maintain close ties with rural educational campuses, government officials working on rural development, and community leaders. We offer internships to students of well known institutes and provide them access to real technical problems of real communities.

Why research on robust networks for rural areas?
An important issue in disseminating information to users in the developing world is that they operate in a resource constrained environment. Power outages, equipment failure, low bandwidth and unreliable network connections are some of the reasons that it is difficult to maintain Quality of Service over these networks. With the rapid growth and spread of the Internet, bandwidth and infrastructure requirements are increasingly under stress. This is more so in rural areas that suffer because they depend on the weakest links in the chain which are highly vulnerable to failure. Enabling and enhancing access and providing end users with more useful services across a broad array of devices and their computing resources (including thin, wireless clients) in an efficient and user-friendly manner is critical.

We plan to address these issues by developing the necessary solution and deploying them in rural parts of Maharashtra and then publishing our solutions for the benefit of other organizations and countries.

For networks to be robust:

• Weakest communication link should not be the determinant in the access to information.
• Total dependency of users on network infrastructure must be reduced (as much as possible)

This requires that for each consumer of information we must determine the quality of service received and identify the weakest links in their individual view of the network. This information is utilized to plan for redundancies and backup mechanisms so as to equip each information delivery point with the tools to deal with failures. These technologies ensure that the end user receives a minimum level of service even in adverse conditions. 802.11-based wireless networks are also envisaged as a cheap alternative to fixed line or cellular network access to Internet.

Why did we choose Pabal as our pilot site?

Vigyan Ashram is an educational organization that caters to school dropouts and focuses on teaching practical skills that will allow students to start their own businesses when they leave school. Situated in a dry, hilly region, in a village called Pabal, Vigyan Ashram - whose primary focus is rural education - is running an Internet Service in an effort to become self-sustainable.

What are the services currently provided at Pabal?

Pabal is connected via 64 kbps leased line. Pabal kiosks are connected to the leased line using wireless corDECT technology developed at IIT Madras [2,8] that has been deployed successfully in many parts of rural India. The corDECT implementation allows quick installation (1 day) of a new connection in a circular area (25km radius) around their 45m tower and Pabal has currently given out over 70 internet connections. 26 of these connections are being run on a single-computer cyber-café model (kiosk) and are run by local entrepreneurs. The cost of an internet connection is comparable to what one pays in urban areas. Typical download speeds vary between 16 and 32 Kilobits per second permitting Internet browsing, Email, Image transfer, Downloading music etc. Some of the popular uses include:

1. Interactive educational CDs
2. Examination results
3. Resume writing and Job applications
4. Email
5. Entertainment - music, photos, contests
6. Phone calls within the local loop

Each kiosk has a multimedia PC, Printer, web camera and a wall set for internet connectivity. A rapidly expanding net-savvy group of youngsters in Pabal are using the internet for communication and entertainment. On the other hand, business related uses are still to be seen.

What are the problems being faced at Pabal?

Problems encountered in Pabal are typical for ISPs in rural areas. They include:

1. Lack of dependable leased line, power and reliable queueing. For example, locals in the area were not able to check the results of Class XII board exams on the internet
2. Lack of confidence in services since they're not reliable (e.g. Examination results were not accessible due to leased line failure).
3. Unreliable leased line (the only gateway to the internet) at Pabal
4. Daily 4 hour power cut at Pabal; generator keeps the tower functioning
5. Unreliable power supply at kiosk
6. Occasional failures in corDECT servers
7. Low bandwidth of less than 2 Kilobytes per seconds at times
8. Complete dependence on the internet for services

What solution do we propose?

Our objective is to deploy a wireless point to point link using the IEEE 802.11 protocol to link the access centres in Rajgurunagar and Pabal. The Internet service at Pabal is experiencing frequent breakdowns in its leased line. By providing a reasonably high bandwidth link between the centers, both centers can continue to function even if one of the leased lines is down. We will use existing towers at Pabal, Retawdi and Rajgurunagar.

These locations are separated by 20 kilometers and we plan to use 802.11 equipment to achieve speeds exceeding 55 Mbps to link the existing towers at Pabal and Rajgurunagar. This will be done through an existing intermediate tower at Retawdi which enjoys line of sight with both these towers.

Proposed 802.11 Link between the Rajgurunagar (also called Khed) and Pabal towers

How does the proposed link mitigate these problems?

• Prefetch a localized subset of the information pertinent to Pabal from the global internet so that at all times any Pabal kiosk user receives some information of a reasonable QoS.
• Add a wireless link between Pabal and Rajgurunagar so that Pabal continues to receive service in case its only leased line is down (currently this is the situation for about 10 days in a month). By linking the two leased lines, the network no longer depends solely on one leased line. Our assumption here is that the probability that both leased lines will be down simultaneously is very small.
• Reduce dependency (as practically possible) of users on network infrastructure. Investigate alternative ways of service delivery under various conditions including power failures, network downtime, low bandwidth and single point failures.

What are the additional services we plan to implement?
The Media Lab Asia laboratory at IIT Bombay has been working on creating smart yet affordable solutions for the masses in India's villages and towns.

Work done at Media Lab Asia at IIT Bombay relates to the proposed work. Below is a summary of some of the current efforts at the laboratory:

Demo versions of price information and the keyboard are already ready. We plan to demonstrate at least three services that will benefit from our approach. Work on some of these services is already ongoing in our labs. We also plan to involve local youth and Vigyan Ashram students in various community project tasks.

How our implementations will benefit Internet services in Pabal?

The technologies we plan to develop and deploy will ensure that the end user receives a minimum level of service even in adverse conditions. Examples of this would be:

• Email service and Q&A service: Currently all web-based email sent from Pabal is totally dependent on end-to-end internet connectivity. We will implement store and forward abilities to ensure that the user is able to send email to a local server. The user is not asked to wait to come back when the "internet is back again". We will also allow the user to type, send email and ask questions in their own language. The Q&A service will have multilingual and meaning based search capabilities performed over a language independent database. What this means is that a user can type a query in Hindi which will be matched against (say) English documents and the results will be displayed in Hindi.
• Market Price Information: The buyer and seller will receive customized market data to allow them to make informed decisions. The data will be in the local language and is cached locally so that the user receives the last update.
• Local content: Local content will be stored in a distributed architecture making it accessible to all kiosks in the area irrespective of problems with the leased lines.

What are the other services possible with higher bandwidth?

More useful services will be possible if the current network is upgraded to one of higher bandwidth. One of the benefits of establishing a long distance wireless network is that we can use this to test services in communities that lie around the towers using high bandwidth (50 Mbps upwards) connections. Possible services:

1. Distance education: Vigyan Ashram currently conducts its distance courses via CDs sent over postal mail. We can help use our current experiences in establishing distance education program in over 10 campuses to assist Vigyan Ashram evolve its non-formal educational program.
2. Medical diagnostics / consultation: Linking Pabal with Rajgurunagar over a reasonably high bandwidth line is the first step in connecting primary health centers in Pabal village region (over 5000 families) to the good medical facilities provided in towns (such as Rajgurunagar) and cities (as Bombay).
3. Cable TV service over IP: In our field surveys we have identified cable TV as an extremely popular service. We will work with cable providers in Rajgurunagar to extend this services to areas now covered in the Pabal region.

Project objectives

6. Provide a localized subset of the information pertinent to Pabal from the global internet at all times locally so that any Pabal kiosk user always receives some information of a reasonable QoS.
7. Add a wireless link between Pabal and Rajgurunagar so that Pabal continues to receive service in case its only leased line is down (currently this is the situation for about 10 days in a month). By linking the two leased lines, the network no longer depends solely on one leased line. Our assumption here is that the probability that both leased lines will be down simultaneously is very small.
8. Reduce dependency (as practically possible) of users on network infrastructure. Investigate alternative ways of service delivery under various conditions including power failures, network downtime, low bandwidth and single point failures.
9. Demonstrate at least three services that will show the benefits of our approach. Work on some of these services is already ongoing in our labs.
10. Involve local youth and Vigyan Ashram students in various project tasks

Project beneficiaries

• 26 computer kiosks being operated by rural entrepreneurs and their users in and around Pabal
• Vigyan Ashram, an NGO that has been working in the field of rural vocational education for 24 years will get a reliable internet connection to run its ISP services.
• Local youth get experience in surveying and deploying internet-kiosk based services.
• Our work will be published on the internet and in leading journals and other ICT interventions that face problems of low bandwidth and unreliable network connections can learn from our experiences.

Project sustainability

• The Wireless ISP service and kiosks are already close to achieving self-sustainability. Revenues are being generated from the services mentioned in 6.2.
• We have employed 2 people for this project at Vigyan Ashram through funds from IIT Bombay. They will continue to remain on rolls as employees of IIT Bombay for at least another 3 years.
• We will be opening a Project office at IIT Bombay for this project that will continue to work on field deployments in other regions of India and Asia that will be funded in part from project funds obtained from IIT Bombay.
  

Project methodology
The objective is to provide a localized subset of the global internet at all times locally so that the user always receives some information of a reasonable QoS. Single point failures in the network result in graceful degradation of service as opposed to its termination. This will be accomplished by creating community caches of sites accessed by users in the region.

Wireless corDECT technology has been deployed by n-Logue communications in India in more than 10 sites in the country. Our research will initially focus on their site in Pabal and investigate how we can use WiFi to make their network more robust. Our initial investigation shows that corDECT-WiFi hybrids will allow Wireless ISPs to increase coverage area while keeping the costs low.

Smart distributed caching and store and forward methods will enhance the quality of services being provided in the Pabal area. These will be utilized by our multilingual query tracking system and price information system.

Steps to be taken

1. Assuring guaranteed QoS to user (both kiosk operator and farmer)
2. Building redundancy in the network
3. Maintaining history of usage
4. Utilizing usage patterns to provide basic services
5. Using history of accesses to create local caches
6. Converting dynamic pages into static pages for local cache
7. Compression techniques for images and video
8. Capacity building of kiosks so that they transform from computer operators to agents who can get day-to-day work done for the locals
9. Identify and implement services that can be delivered without resorting to a live internet connection

Project Schedule

Project outputs

1. Report on field study of Pabal's network.
2. Operational long distance WiFi link between Rajgurunagar and Pabal
3. Localized website for market price information
4. Content pages of the villages in the Pabal region
5. Multilingual Question and Answer engine
6. Report on kiosk services before and after deployment

Project monitoring

1. Simulations conducted at our labs in IIT, Bombay
2. Daily feedback from field engineers based in Pabal
3. Feedback from kiosk users (kiosk operators and locals)
4. Online portal for the project for publishing results

Project Evaluation

1. Quarterly reports on progress
2. Final report on lessons learnt from the deployments and list of new sites identified
3. Independent half-yearly evaluation by Vigyan Ashram
4. Independent annual evaluation by the International Development Research Centre