The south Asian region has had a telecommunications infrastructure since the mid-nineteenth century. The concept was introduced formally in 1853, although it may only have been implemented in 1856. This period marked the ending of the suzerainty of a private organisation, the East India Company, over much of the land area of this region. The British government formally took over direct control subsequently, and an act of parliament was passed in 1885. This legislation, the Telegraph Act of 1885, unfortunately still operates in the region, although it now consists of many independent countries, four of which were represented at the workshop.

Since it was originally built to support a colonial regime, and its subsequent growth and management reflect this, until the mid-20th century, when Britain handed over governance of what is now largely Bangladesh, India and Pakistan. At that time, postal services or their offshoots mainly handled control of telecommunications. With the increasing ubiquity of the telephone, this segment has tended to gain distinctly separate organisational control.

Most public two-way access to the medium is through wire-line communications, although radio has been used in other countries effectively. This is because two-way radio has tended to remain largely the preserve of the military services. Audio broadcasting through radio frequencies has been extremely popular and so has television, when it grew through the 1970s and 1980s. However, control of these media, just as in the case of two-way communication, has remained largely in the hands of the government, until the advent of satellite broadcasting and cable distribution.

The use of radio for two-way communication remains in its infancy in this region, mainly because radio frequencies, originally generously allocated for military use, are highly restricted for public use. The bandwidths being used in other countries for cellular and satellite telephony are apt to run into conflict with frequencies in use in these regions. The massive growth of satellite-based television broadcasting has meant that control and use of the airwaves has become much more important than before, but change is slow.

The Internet (created in the early 1960s) is a two-way telecommunications application with a difference: it uses devices � computers � endowed with a great deal of complexity, more than any previous such devices, to receive and transmit pure digital signals. The signals blend messaging with controls in a single multiplexed data stream. Since 1993, the Internet has been enhanced with a new development, namely, the creation of databases that users can access without mediation. Coupled with very easy to use graphically designed computer applications � browsers � this has made the medium available to millions of people who do not need much more than a basic command of written language to become adept.

In actual fact, the growth of the Internet globally has outstripped any previous innovation by orders of magnitude. If 50 million is taken as a measure of the number of users needed to make a technology ubiquitous, then the automobile took some 30-odd years to reach this level and television twenty. The Internet, however, has taken only five, and is well on its way to doubling that number in less than two years more.

Its greatest advantage is that it needs no new infrastructure to get started, since it can be carried over ordinary telephone lines. It does, however, demand a complete revamping of the way that infrastructure is used, and this is the source of much of the frustration experienced by those who champion its use for development in emerging economies, especially the countries of the south Asian subcontinent and its neighbourhood.

Over the years, telephone engineers have determined a simple rule of thumb for building telephone network infrastructure, which is that the average telephone call lasts for three minutes. Since an analogue telephone call demands the total allocation of a point-to-point circuit for completion, the carrying capacity of a telephone network is a linear function of the number of circuits available over a given period of time. This logic is extended to cover the average carrying capacity, measured over a full 24-hour day, or the peak capacity, measured whichever time of day the maximum number of calls is attempted. The latter is termed the busy hour call attempted' capacity, and is the real measure of capability of an analogue network.

As telephone infrastructure grows, engineers develop techniques to enhance the peak capacity without necessarily adding complete point-to-point capability. For instance, a star network, where telephone switches are connected to each other over many paths, naturally has more capacity than a linear network, where the ability to complete a call at that moment is dependent on each switch being available in the singular path. This is certainly cheaper than attempting to wire as many interconnect circuits (circuits between switches) as there are telephones. Further increase in effective capacity (i�e, in availability) comes through connecting the switches in a mesh network, which enables a circuit to be connected between any two points through any series of switches.

The Internet, however, does not follow the three-minute rule. It is actually creating a completely different usage paradigm, one where the devices remain connected for twenty and thirty minutes at a time. In fact, experts are of the opinion that at some point in the future, people will have devices permanently connected to data lines. Obviously, this needs a totally different approach to both the building of infrastructure and the way that customers, or users, pay for it. It also demands comparatively superior telephone technology, as the earlier generation of equipment produces noise levels that are unacceptable for high-throughput digital messaging.

The growth of the telecommunications infrastructure in south Asia has not been demand-driven, unlike that of the countries where the infrastructure has been built and services operated by private investors. It has been almost entirely investment-driven, dependent on the priority level possible for the allocation of funds from limited public resources.

Before entering into a discussion of the existing telecommunications capabilities in the region, a quick look at geopolitical realities and current technologies will help to put the position in context.

Geopolitical Realities

This part of Asia is usually known as the subcontinent, since it covers an area comparable to Europe or Antartica, the smallest continents. It stretches from Afghanistan in the west to India in the east, with Bangladesh and Nepal placed towards the northern and eastern parts of this area. Other (smaller) countries in this region are Bhutan in the north and Sri Lanka, an island just south of India.

Two significant and dominant features of this area are the Indian Ocean to the south and the Himalayan mountain ranges to the north. Giant rivers also cross the area, the Indus in the west, the Brahmaputra to the north and east, the Ganga and many others. Of these physical features, mountains are common to all the countries, but rivers and the sea are major factors for Pakistan, India and Bangladesh. The last, particularly, has its entire economy (and roughly half its land area) dominated by a river delta fed by both the Ganga and the Brahmaputra. The eastern seaboard is lashed by seasonal typhoons, which repeatedly wreak havoc in the low-lying areas of Bangladesh and India, making it�difficult and expensive to maintain modern infrastructure. Nepal is a mountain country, contained entirely within the Himalayas.2

Politically, these countries were generally under the suzerainty of the United Kingdom for almost two hundred years, but gained complete independence in 1947. The main political area, India, was subdivided into two, and the smaller portion, Pakistan, which had two discontiguous regions in the east and west, subsequently split into two countries with the formation of Bangladesh in the east in 1971. Except for Nepal, there has been a history of armed conflict in the area.

India is often called the world's largest democracy, having maintained a parliamentary system since independence, except for a brief interlude in 1975 when normalcy was suspended for two years. The president of the country does not play an executive role. Pakistan and Bangladesh, which have presidential systems, have both had histories of military dominance, but this has been to a much lesser extent in recent years. Nepal is a kingdom, which like many other historical kingdoms is making a transition towards a more democratic form of effective people's representation.

Technology

Internet connections tend to be of long duration, unlike voice connectivity, which has traditionally been based on the three minutes per call rule. Although Internet usage per capita is still extremely low, this factor has put a strain on the existing investments in infrastructure right across the region, both for local and for international circuit capacity. This problem is quite independent of the technology issue, involving the use of modern digital switching as against the historically cheaper electromechanical switchgear, now obsolete. Data access circuits are far more reliable when they are digitally switched, and so is line throughput. At least in this decade, all investments in the region have only been in digital switches; hence this problem is being diminished naturally. (See the individual country telecom tables at the end).

However, that still leaves the problem of bandwidth open. This is directly related to the amount of physical connectivity between geographies. In turn, physical connectivity is of two kinds: cable and wireless.

Cable again includes two technologies, copper (and to a much lesser extent, the more inefficient aluminium) and fibre-optic. Copper is the traditional cable choice, and a huge infrastructure exists within the region and internationally. It can be laid in the ground or carried on overhead towers, both of which have their own advantages and disadvantages. Fibre-optic cable uses strands of internally reflecting glass fibre tubes to carry modulated laser signals. It has a much greater capacity and is particularly suited to digital signalling, and is also more efficient in terms of material content. Finally, it is more difficult to illicitly intercept signals, hence more secure. All major cable projects in this decade are fibre-optic, and it is unlikely that copper will be used again for long-distance connectivity.

Wireless is also of two types, terrestrial and satellite. Terrestrial signalling, in the microwave frequencies, has the advantage of minimising disruption to existing artefacts (agricultural, traffic-ways or conurbations) along the lines of access, and also to a large extent of optimising the real carriage length, since radio-waves operate along line-of-sight, hence repeater stations can be placed almost in a straight line between places. However, to some extent, technical considerations of bandwidth choice due to the quantum of signal loss over distance means climatic conditions (fog and rain) do cause occasional degradation and quality problems for this methodology. In short, the frequencies commonly used are prone to atmospheric interference.

Satellite wireless uses radio repeaters in space orbit to convey signals between geographies. Public services have traditionally used geo-stationary orbit, which means that the devices appear to hover in the same spatial location relative to any place on the ground, but several upcoming schemes use numbers of mobile satellites for this purpose. One of these, Iridium, has already begun voice services globally. Others plan to offer both voice and data services.

There is an enormous capacity available for geo-stationary satellite signalling, although there are fears that the slots' for future infrastructural expansion are running out. Signalling is built in as a feature of almost all television satellites, which typically host a number of radio switching devices called transponders, each one tuned for optimal performance in a particular bandwidth, some suited to data connectivity and some for television. Satellite communication is usually done in frequency ranges where visual atmospheric disturbances are transparent, but on the other hand, the total capacities per channel are very limited, compared to terrestrial transmission.

Domestic telecommunications connectivity has been mainly microwave over long distances, with wireline networks in the local networks. This is largely due to the fact that the telecommunication companies (telcos) tended to protect their monopolies by avoiding relationships with other agencies possessing access right-of-way. In recent years, there has been some change in this position. Now Bangladesh has built substantial fibre-optic links along its railway tracks and India has recently permitted its railway and power companies to do the same. Nepal, whose terrain and existing infrastructure is not suited to this solution, has decided to go ahead with VSAT connectivity (so has Bhutan).

Teledensity

Teledensity is very low throughout the countries of South Asia (specific country-wise data is in the tables at the end Profiles of States, 1997, Centre for Monitoring the Indian Economy, Mumbai). In Bangladesh (see Tables 1-6), it stands at around 0.4 lines per 100. This is recognised and targets for growth have been set, 1 per 100 by the year 2000 and 4 per 100 by 2010. India is the largest country in the region and in�some respects the most developed. However, there are huge disparities extant within the country, and this is evidenced in the uneven distribution of telecommunications access. Connectivity remains low and unevenly distributed, geographically speaking, from as low as 0.49 lines per 100 (less than one line per 200 population) in Orissa, with an estimated population of 32.74 million in 1995/96, to 2.07 per 100 in Gujarat, whose population the same year is estimated at 42.87 million [CMIE 1997]. India, which is politically a federal republic, has 28 states with a total population currently estimated at over 980 million. Of these, at least 10 have mountainous territory, and several others are significantly hilly, i e, containing areas with elevations 1,000 to 2,000 meters above sea level.

In India (see Tables 7-8) cities are mainly wired using traditional copper (actually a mix of aluminium and copper) cable, but now fibre-optic has replaced the switch interconnects in the major cities. City interconnects have been mainly through microwave links, but an effort is being made now to install a nationwide trunk fibre-optic cable network. This is expected to be in two parts: a cross network running east-west and north-south, and a coastal network linking the eastern and western coastlines. Most of the switches installed in the past used electromechanical technology, but for the past five years or more, only digital switches have been used for capacity enhancement and upgradation. Overall the switches remain mixed.

India is the only country in the region with a reasonably high level of engineering hardware and software technology. In the 1980s, an effort was made to develop indigenous digital telephone switch technology, using a modular approach in order to produce scalable products. Unfortunately, this effort was abandoned in the early 1990s, mainly due to political reasons.

Nepal is largely rural in nature, with 90 per cent of the population living outside the main cities. The biggest city, the capital Kathmandu, has 2.7 per cent of the country's population. Unfortunately, Nepal also has the lowest road length in relation to area or population among all countries, which coupled with poor communications infrastructure means that many villages are isolated from the rest of the country. Barely a quarter of the village councils (village developmental committees) have telephones (as mid-1998) (Tables 9-10), but the government has now decided that telecommunications are a priority. An overall tele-density of just 0.9 lines per 100 people represents a doubling from two years before, when there were less than 1,00,000 phones in the entire country. The figure is expected to more than double again in the two years till end-2000, raising it to 5,00,000 lines, or around two lines per 100 persons. On the other hand, the entire circuit is digital and all telephones have access to international subscriber dialling to 131 countries.

Pakistan also has a fairly modern domestic service (Table 12), but connectivity is uneven, with the worst service levels being in the north-eastern regions. This area is not only under semi-military rule, even the local telephone services are managed by the military, and village level connectivity is very poor indeed, well below the national average of 2.2 per hundred.

Physical infrastructure being a limitation, with the country's long route transmission systems mainly composed of microwave, UHF and VHF radio links, Bangladesh is now encouraging the growth of mobile telephony, which is entirely run by the private sector. The railways have built a fibre-optic link along much of the track length, which could help, but the modalities of public access have not yet been determined. Most new switches are digital, but there is still a legacy of electromechanical switchgear. The long-range plan is to replace all analogue circuits by digital by the year 2005. The most important expansion and rehabilitation programme being taken up is the laying of 12 core Optical Fibre Cable between Dhaka and Chittagong, which is the busiest route and still analogue.

Nepal has decided to move towards extensive VSAT tele-connectivity to alleviate the problems of poor traditional infrastructure, which prevents them from the kind of piggybacking that other countries are now examining. Pakistan is faced with a problem created by the early liberalising of Internet access, which has put a heavy load on a small network.

International Connectivity

India is the only country in the region with a clearly distinct (government owned) agency for the provision of international connectivity, Videsh Sanchar Nigam, which was also the first licensee for the provision for public Internet access (i e, an ISP). As a result of this monopoly, and not having to subsidise local connectivity from its revenues, it has been an extremely profitable company, with a strong programme for investment in expansion of infrastructure, which accelerated in this decade. It has been allowed to retain the exclusive arrangement on international (since the end of 1998, only for voice) connectivity till the year 2004. Data services, especially Internet access, have been opened up, but the guidelines (rules) for privately operated gateways had not been published till middle of July, 1999.

Currently, only the Software Technology Parks of India (STPI), a government owned industrial park operator, has such a facility for its lessees. Although over two hundred ISPs have been licensed, only two national privately-operated services have begun so far, and they will very likely opt to open their own gateways.

Nepal gained access to the Internet with the help of ERNET, India's education and research network. This limited arrangement, which initially offered only shell account access, was later routed through Singapore and now Australia. Voice telephony is still routed through India, the logical path. Bangladesh and Pakistan also have very limited international connectivity, and this may be attributed to the fact that all three countries each use a single government agency for both domestic and international services. Investments in international pathways have been delayed for short-term economic and political reasons. Pakistan has now negotiated a link to SEA-ME-WE III, a cable that runs from Europe to Asia.

Internet Services

All the countries in the region have liberalised Internet access in one way or the other. However, this has not been a smooth process, and there is considerable resistance from the erstwhile government�owned telecom monopolies, who are�accustomed to exercising complete control over all forms of tele-connectivity.�The result is a great deal of resentment on both sides, with the ISP's bewailing the lack of co-operation that endangers their relatively under-capitalised businesses.

India was the last country to liberalise Internet access, which was till the end of 1998 monopolised by VSNL, citing its monopoly on international connectivity. In the other countries, private ISP's were allowed to set up operations from 1995 onwards. In Nepal, the first Internet access came about as a result of a co-operative move from India's ERNET, which extended connectivity to Nepal for limited shell access and e-mail. ISP's in the country later set up private arrangements.

The Internet service business in Bangladesh has been privately run until recently, with two major ISPs and several smaller ones. With local access to international telephony bandwidth being difficult, these services have been almost exclusively run�on VSATs (Very Small Aperture Terminals, a radio-satellite technology). However, Bangladesh Telephone and Telegraph Board (BTTB) has been lobbying for permission to offer Internet services. Efforts to organise a country Internet exchange have not fructified.

As mentioned earlier, the history of the region's telecommunications infrastructure predicated monopolistic tendencies. While all the countries have attempted to deregulate and liberalise their data and voice services, encouraging private investment and expanding the range of offerings, this process has been unevenly implemented. There appears to be insufficient comprehension of the potential of voice and data communications to be a key enabler in societal transformation, both economic and social.

Regulations

All the countries are still ruled by the Telegraph Act of 1885, with minor individual modifications made from time to time. While there have been many public debates on the need for sweeping changes in this area, little action has emerged. Ad-hoc changes, mostly only loosely supported by the Act, have led to long-drawn out and expensive lawsuits.

Control

Telecommunications in Bangladesh are principally the responsibility of the BTTB, but there are two other, smaller, private service providers for basic telephony. However, their growth has been stunted and they have little impact on the scenario. There is no independent regulator in place, but attempts are under way to set this up in the form of an independent Telecommunications Regulatory Commission, according to an Act that is currently prepared for tabling.

The Nepal Telecommunications Authority (NTA), a body formed in April 1998, now regulates both telecommunications and the Internet for that country. In November the same year, the government liberalised a number of services to the private sector, in the hope that private investments will bolster government initiatives, particularly in the field of VSAT linkages. VSAT technology is currently the only feasible method to connect isolated and remote villages electronically. Neither the tariff nor conditions have been set for licences to run VSAT links. Till this time, only the government-owned Nepal Telecommunications Corporation (NTC) had the charter to provide telecommunications. Previous attempts to introduce competition in 1996 had failed. NTC still holds a monopoly on providing fixed (basic) network services.

India set up the Telecommunications Regulatory Authority of India (TRAI) in 1997, but it has been an extremely poor agent of change insofar as effectiveness is concerned. Public wrangling and lawsuits have marked its existence, with its ability to enforce rulings particularly weak. Of late, the government (in the form of the prime minister's Office) has further weakened it by creating and attempting to implement a new telecom policy intended to remove barriers in the way of expansion of private networks, without reference to TRAI.

India has liberalised the provision of Internet services since end-1998. Till mid-July, however, private companies had to buy international connectivity services from VSNL, which competes in the business. Guidelines for private gateways have now been announced, but no one has been able to start as of yet. Suspicion and outmoded thinking dominate the transfer of adequate rights to private players. For instance, ISP's with private gateways will be required to cache all traffic and turn over to the government's investigative agencies information pertaining to any and all potentially criminal activities, on demand.

Three types of ISP licenses exist, all being almost free for the first five years. However, providers must furnish bank guarantees for the fee amount, Rs 2 crore for national providers, Rs 20 lakh for providers in certain major cities and Rs�3 lakh for less important markets. While a number of services have started, quality of connectivity is reportedly lower than that of VSNL. However, they have introduced a number of innovations to help novice users to connect. The total number of connections has reportedly doubled to 400,000, with VSNL's share of this market being 50 per cent.

Basic telephone services in Pakistan, domestic and international, are completely under the government's control, through the Pakistan Telecommunications Company (PTCL). In the mountainous northern sector, the army's Special Communication�Organisation (SCO) runs the telecommunications service, and the service is reportedly very poor, reaching only a few of the estimated one million population. However, Internet services are open and there are a large number of � over 50 � service providers in the other parts of the country.

The Pakistan Telecommunication Authority (PTA, http://www.pta gov pk/) regulates tariffs and services, licensing and the use of radio-frequency spectrum. The Internet is licensed separately in a category called Electronic Information Services, which also includes electronic commerce initiatives. The government has an open policy for data communications and Internet, to encourage and spread its use in Pakistan with the help of the private sector. There are three categories of licence for the Internet, a Big Citywide Service (Karachi, Lahore, Faisalabad, Islamabad) costing Rs 3,00,000, a small citywide service at Rs 1,00,000, and a countrywide service at Rs 5,00,000. There are further fees, including an annual renewal fee equivalent to 20 per cent of the initial licence fee and 4 per cent of annual gross revenue as royalty to the PTA. For fledgling services with expensive and poor quality connectivity, this can be crippling.

Internet Domains

Generally speaking, it has been difficult for the various private and public sector organisations providing telecom and Internet services in Bangladesh to work together on any initiatives. The country top-level domain, .bd, remains without administration and is virtually unused as a result. India's top-level domain, .in, is administered by the National Centre for Software Technology, (http://www.ncst.ernet.in), which also runs ERNET. One of the ways in which it has attempted to control use of the top-level domain is to insist that all registered users must host their websites in India. Since the pricing and quality of service within the country is not competitive, especially compared to the US, there has been an exodus of service-oriented sites. In addition, while the registration service is technically free, in actuality documents must be submitted through the site service provider, who then charges for this service. However, there have been relatively few complaints of misuse of brand and company names, unlike the confusion and arbitraging of names that has taken place in other top domains, especially the countryless' (dominated by US sites) domain.

Mercantile Communications, a leading ISP, and the Australian company ISP Connect jointly administer the.np country level domain in Nepal. The top-level domain for Pakistan is pk, which is centrally administered by the Pakistan Network Information Centre (PKNIC, http://pknic.net.pk/), with the co-delegate assistance of a number of Internet companies. A committee that includes members from the ISP and IT communities sets PKNIC policy.

Based on the discussions at the South Asian Internet Workshop in Dhaka 1999, several policy recommendations can be made to policy-makers and Internet professionals to further strengthen the Telecom Sector in South Asia. Set out under three categories these are as in the Table�A.