|
|
Analyses |
|
| About | Documentation | Events | Lists | Subprojects | Contribute | Mirrors | Site Map |
| Design Axes for the Indian Language Computing Market | ||
|---|---|---|
| Prev | Next | |
In this section we examine a few initiatives that have attempted to provide solutions for the computing needs of the Indian subcontinent.
Solutions attempting to build an e-governance infrastructure based on the ubiquitous IBM PC architecture are examined in Section 3.1. Section 3.2 examines the Simputer, an indigenously designed handheld device. Section 3.3 examines the recent explosion in cellular phone connectivity.
The (IBM) PC architecture is by far the most popular computing platform in India on account of its comparative cost-effectiveness. A number of initiatives have attempted to use PCs as a building block for Indian language computing infrastructure--for example, many state governments are building e-governance systems in local languages, using PCs running Microsoft operating systems as web clients.
E-governance involves centralized data (possibly from government registries) being made available to the general public via ``e-governance kiosks''. These kiosks allow Indian citizens to request services, request information, submit personal data etc. in their local language. Kiosk owners provide the infrastructure required to access e-governance data for a fee to their customers.
Barring a very small number of open-source enthusiasts, the vast majority of PC users in India use the Microsoft Windows(R) operating systems. Versions of the Windows(R) operating system starting with Windows 2000(TM) have had built in support for a small number of Indian languages. Hindi, written using the Devanagari script is perhaps the best supported language under this OS.
Analysis
A typical E-governance kiosk would require a high-end PC, a printer and additional support infrastructure like internet connectivity, lighting, and backup power. The power requirements of the whole system are high, with the PC itself requiring about 300 Watts of power.
Thus it would be difficult for every village in India today to have an e-governance kiosk. The market penetration of such kiosks would limited to areas with a functional electric supply like India's cities and small towns.
Indian language input in this system is using the PC's keyboard. As noted earlier, using a 100+ key keyboard for Indian language input raises the learning curve for new users significantly.
Note: Windows(R) supports the Inscript keyboard layout. While this keyboard layout works well for Hindi input, it has been reported to be less than optimal for other languages.
The base operating system in this solution uses Unicode for the underlying character encoding. This works well enough for the Devanagari script, there are pending issues with most other Indian scripts that preclude a 100% solution being offered.
While the e-governance system is administered centrally, with its data on a central server farm, the information content it carries is local in character. The Bhoomi project of the Karnataka land administration allows land records to be updated from sub-district level data centers. This effort thus supported a limited form of local information gathering.
The value provided by the e-governance kiosk lies in the transparency it brings into the working of the government (i.e. no bribes) and in the streamlining of cumbersome government procedures for common tasks.
This effort does not directly address social policy issues. However, since it requires physical access to the e-governance booth it is suspectible to being ``choked'' if attempted in a rural context. In practice though, the high power requirements for these kiosks would restrict them to small towns and urban areas where the hold of the Indian caste system is not as strong.
The Windows(R) operating system enjoys a very high level of support in the world-wide developer community. The Indian language aspects of the OS are however, less well known. This drawback, however, does not affect the project as most of the development work would be done at the server end, with the end users only needing to know how to use a web browser.
In summary, the power hungry nature of a PC configuration needed to run Microsoft operating systems is the biggest drawback to the use of these operating systems as a foundation for Indian language computing. The lack of an intuitive way of interacting with the computer in Indian languages is its second biggest obstacle.
The Simputer is a device that is very similar in features to existing personal digital assistants (see the sidebar Simputer Features). The unique feature of the device is a smart-card interface that (potentially) allows a single Simputer device to be shared by multiple users--user specific data would reside on a cheap smart-card.
|
Simputer Features
|
The Simputer project has attempted to make it possible for end-users to share a single device by allowing the possibility of per-user data residing on a smart card. This would allow, say, a village community to share the cost of a Simputer device--helping defray the rather high initial cost of the device.
However, the design of the Simputer does not directly address the core issues in Indian language computing:
Analysis
Early prototypes of the Simputer were battery powered using 3 AAA cells, but used to offer only a few hours of battery life. Later models tried to alleviate this problem by offering an add-on main adapter kit, and by adding support for re-chargeable batteries. As it stands though, the power usage of the base Simputer is too high for it to be effective in the Indian context.
Designing a Simputer-like device with radically lower power consumption is possible, but difficult. For one, such a design would be a departure from the reference hardware design that was adopted for the base Simputer. Secondly, and perhaps more difficult, low power consumption requires fine-grained power control; this would need the Simputer's operating system (a near-stock Linux kernel) to be radically changed.
The Simputer today offers an on-screen keyboard for US-English input and has an experimental tap-a-tap input method also for US-English input. Text to speech output is possible in a few Indian languages. Entering data in Indian scripts however, is difficult.
Since the Simputer offers a touch screen it could potentially offer its users a direct way of entering text in Indian scripts with the system performing handwriting recognition. The requisite software support is however, missing.
Since the Simputer's operating system uses existing open-source components, its ability to process data in a standards compliant way is dependent on the support for Indian language standards in the open-source world. This support, is today still a work in progress.
Once its basic issues of power, usability and interoperability are solved, the Simputer function as a data collection unit for a system designed around local information sources.
The initial applications of the Simputer appeared to lack significant value addition for the rural Indian consumer. For example, attempting to replace a paper ledger costing INR 30 with an INR 30,000 Simputer to do accounts is clearly not an economically feasible alternative.
The true value of a device like the Simputer would be evident when we have networks of local information in place. Till this time, it would be difficult to design a ``killer application'' for the Indian market around the Simputer.
While the addition of the smart-card interface can potentially bring down the entry costs for a user wishing to use the device, existing social systems in India can still prevent substantial market penetration of this device. Given the existing caste prejudices in the typical village, it is difficult to see a village headman and an untouchable sharing the same Simputer.
The Simputer hardware license encourages experimentation; however, few commercial and academic entities in India can afford the initial layout needed to do the necessary hardware experimentation. Experimentation at the hardware level is certainly beyond the reach of an individual.
On the software side, nearly all of the software bundled with the Simputer has been derived from the worldwide open-source community's work. While the worldwide open-source community is vibrant, there are very few open-source developers interested in Indian language work.
This puts the Simputer effort in a peculiar situation: changing the hardware to better suit Indian conditions would result in an increase in the effort to port and track ongoing open-source development. At the same time, the low market potential for the existing devices makes it difficult for the platform to gain market momentum.
In the recent years, cell phone usage in India has exploded. In just a few years, the installed base of cell-phones has exceeded that of computers and this growth is not showing signs of slowing.
Cell phones are fairly sophisticated computing devices, containing computational abilities exceeding that of a 1960s mainframe. The typical mobile phone has inside it a 32bit ARM or Hitachi SH family processor, a few megabytes of memory and a small (a few hundred pixel wide) display.
We now examine how the cell phone fares on our seven axes:
Analysis
Cellular phones are very low power devices. They can operate for hours on a single battery charge and can be recharged from a number of sources, including the electric power available inside automobiles. They are thus usable even in remote, power-starved villages.
The cell phone network's base stations require much more power to operate and so need to use power from the electric grid and from alternative sources like fuel-based generators.
For the task that they are designed for, cell phones are eminently usable. You type in a number and the phone makes a call.
Complications soon arise, when the cellular phone is used to send text (e.g. when using the popular Short Message Service). The numeric keypad is difficult to use for English text itself: for Indic scripts, with their numerous characters, and complex transformational rules, the numeric keyboard is singularly inefficient as an input device.
The cellular phone's display tends to be small. Enabling such a display to present a meaningful amount of Indian language text is a challenge.
Although voice data (and to a lesser extent multimedia data) can be readily exchanged among cellular phone users, Indian language text interchange using the cellular phone still has many open issues. For example, while a few cell service providers have announced the availability of text messaging in Indian languages, it is not known whether these services use standard character encodings. The proprietary nature of the cellular phone platform could be a barrier to achieving interoperability.
Cell phones are a good example of the attractiveness of (contextually) local information For example, even when using voice, personal calls tend to be made between friends and relatives, while businesses tend to call their suppliers and customers.
Some cellular service providers offer a limited form of contextually sensitive data dissemination (e.g.- weather forecasts). However, the information flow in such systems is fundamentally one way (from the service provider to the user). We are not aware of any system where the cell user is a provider of data, although there are no hard technical barriers preventing such a system from being built over existing cell networks.
The ability to communicate easily is perhaps the most important driver behind the explosive growth of India's cellular networks.
The ability to share data in local languages (in addition to voice today) would be the next great value addition for the local market.
The cell phone by itself, like the ubiquitous rubber sandal is not inherently ``choke-able''. Rather, its very nature (wireless, low-power operation) allows it to bypass the existing social structures in Indian society.
The administrative side of cellular networks (allocation of numbers, billing etc) is vulnerable to being ``choked'' by social and governmental pressures. It is not known to the author whether such market inefficiencies are present in today's cellular networks, and if so, to what degree.
The cellular phone is not well known as a development platform.
Firstly, the architecture of most cellular phones is proprietary: a handful of manufacturers make most of the world cellular phones. While the communication protocols between the cellular phone and the rest of the telecommunication network are known, the operating systems used inside the cellular phone are nearly always proprietary. Consequently, it is difficult for a developer to develop portable applications for the existing cellular phone system.
There are no open-source cellular phone operating systems that the author is aware of.
Further, an Indian language developer community centered around the cellular phone architecture is missing.
In summary, the low power consumption and ease of use offered by cellular phones have helped the cellular network to reach a far greater number of Indian citizens in just a few years, than computers have managed in decades.
There are many hurdles to be cleared before this platform can be the basis of an ubiquitous Indian language computing infrastructure. First, its usability with Indian language scripts needs to be dramatically improved. Next, the system needs to be designed giving priority to interoperability. Next, more applications that process local, context dependent information need to be developed. For all this to happen, the system architecture needs to be made more open, and its developer community needs to be made stronger.
This, and other project documentation, can be downloaded from [ http://indic-computing.sourceforge.net/documentation.html ].
|
Copyright © 2001--2007 The Indic-Computing Project. Contact: jkoshy |
View document revision history
|