ashok@tenet.res.in

1. Introduction

India today barely has 20 million telephone connections and less than half a million Internet connections for its billion people. With less than 2% teledensity, we can not talk about telecom and IT playing a major role in the nation’s development. Recognising that Internet Access could result in a divide between haves and havenots, deeper than any existing so far, the nation wishes to grow to 100 to 200 million telephone and Internet connections very quickly. That this is not possible without participation of the private sector has been recognised for long.

Till around the end of the seventies, the sole telecom operator in India, Department of Telecommunications (DoT), procured all its equipments from central government controlled companies like ITI and HTL [Vittal]. In early 80's, the liberalisation process began, first by accepting procurement of telecom equipment through state government enterprises. It took another four to five years, before private sector was allowed to manufacture telecom equipment. The late 80's saw a major initiative when franchisee -run long-distance public call offices (STD PCOs) were allowed to be opened. The STD PCOs quickly spread to all parts of the country and today nets about 20% of the total telecom revenue of the country. The process continued, when in mid-92, operation of value-added services were opened up for private organisation.

1994 was the next major turning point. First, the cellular licenses were issued to the private sector. At the same time, the 1994 National Telecom Policy (NTP) laid the foundation of allowing private sector to operate Basic Services (one operator in addition to DOT for each circle or state). This policy document attempted to clearly enunciate the goals of the liberalisation process. Universal service and making reliable telecom available to all sections of population was obviously the goal to aspire for. No other interest was to come in the way to fulfil this goal.

Even though this major historic step was taken in 1994, the subsequent developments left a lot to be desired. The high license fees bid by operators created initially a lot of euphoria and later a lot of confusion. The long-distance service was not opened up as expected and remained a DoT monopoly. The unclear interconnection, spectrum utilisation and right-of-way policies slowed down all efforts. Further, the expected corporatisation of DoT never happened.

Even with these difficulties, the process moved forward. In 1997, establishment of Telecom Regulatory Authority of India (TRAI) was a major gain, though subsequent confusion about its role and power left a lot to be desired. The next major initiative was the 1998 Internet Policy. This freed Internet operation completely, with Internet Service Providers (ISPs) even allowed to establish their own Access Networks.

The year 1999 saw a major initiative from TRAI. For the first time, it talked about the cost of establishing a Telecom network and corresponding prices that provides a minimum return. Over the years there had been a drastic reduction in the cost of setting up a long-distance network, but the enhancement in cost of providing access network had nullified this long-distance cost reduction. TRAI suggested significant reduction in long-distance charges, and simultaneous increase in rentals and local call charges to bring prices closer to the cost. Unfortunately it did not go down well politically [1]. The second major initiative of TRAI was to reduce leased circuit charges which enabled significant development in Internet and Intranet services.

But by 1999, the major problem that faced the nation was due to the fact that the Basic and Cellular licensees were unable to pay the license fees due from them. There was a fear of large-scale collapse. The operators were pleading with the Government to do something, so that the telecom liberalisation process is not totally derailed. It is in this environment that the NTP'99 was born. The key features of NTP'99 are

• license fees for Basic and Cellular operation converted to revenue-sharing plus entry fee
• allow multiple operators in the same circle removing the restriction imposed earlier
• existing operators allowed to migrate to revenue sharing; the payments already by them is to be taken as entry fee for these operators.
• national and international long-distance operation to be opened up in 2000 and 2003 respectively.

It is unfortunate that NTP’99 was born as a reaction to prevent private telecom sector collapse. But even then the NTP'99 is a part of twenty year old process; a process typically consisting of one step forward, new problems resulting therefrom and correction of the resulting problems. The process had and continues to have the aim of providing everyone a telephone and Internet connection. On the whole, the NTP takes us forward. But will NTP'99 take us more rapidly towards the main objective? This paper focuses on the limitations of NTP'99. The analysis is focussed on the Basic Services Operation.

The operators, both mobile as well as Basic Services Operators, had bid very high license fee payable by them in return for duopoly rights. The amount bid was indeed very high — some of the states had a bid in excess Rs.100 billion to be paid for a 15 year license.

It is this high license fee which has made the private telecom operation unviable within
3 to 4 years. Today’s evaluation is that the bid amount was not just a little high — it is believed that the error was of a factor of 5. It is important to understand the reasons for such an error. The high bids were based on

• assumption that India is a very large market with over hundred million middle class customers
• assumption that this large potential subscribers have reasonable ability to pay.

In fact the first assumption was not too far off the mark. Most cellular licensees got more subscribers in the first two years than they had hoped for. But it is the ability to pay, where the major error seem to have been made [2]. India is indeed a large potential market, but with very limited ability to spend. Why did all these operators, with an army of MBAs, foreign experts and foreign partners, not understand this simple fact about India?

One does not want to cry over the past mistakes, but the issue is have we learnt our lesson? Do we now understand the market better and know what to do, or are we going to proceed and make the next major mistake?

The most important aspect of NTP'99 is that the operators will have a Revenue Sharing arrangement rather than pay license fees for operation. This, of course, solves the immediate problem of operator insolvency. But why use revenue-sharing and not just lower the license fees? May be the obvious answer to this question is that it may have been politically suicidal to accept lower license fees.

The more serious implication however is something else. License fee bidding requires that one understands the market fairly well, whereas revenue-sharing does not. The licensor is now willing to share whatever he/she gets as revenue. A good estimate of market is relatively much less important from the licensing point of view, though it may continue to be important for financial closure of the project. The switch from lincense fee bids to revenue sharing is in fact some sort of acknowledgement of inability to accurately estimate the market and in fact points to the uncertainties.

The more obvious question is, what now prevents the private operators to concentrate on providing service to only a few high-paying subscribers? It makes perfect business sense as less than 20% of subscribers today contribute 80% of the revenue. Even if high cost technology and systems are used to provide service to these few, the operators are likely to make a profit. In a fixed license fee regime, as one expands the subscriber base, one is likely to gain (vis-a-vis license fee payment) even with marginal revenue from new subscribers. With revenue sharing, this is not so. Why bother to provide service to the subscriber who are small consumers, at least to begin with? In fact a few of the Basic Service Operators (BSO) today refuse to provide service to all but corporate subscribers — a direct consequence of using high cost imported technology. Is revenue sharing going to encourage this further?

BSOs taking the high-paying subscribers away from DoT could hardly be the goal of the country’s Telecom Policy and the privitisation process [3]. In fact the whole privatisation process started with a promise of Universal Service – providing telephone (and Internet) to ALL.

In the light of this, the removal of Universal Service Obligation for the Basic Services Operators in NTP 99 looks very problematic. A universal service tax (of a few percent) is to replace this Universal Service Obligation. Is this a way of saying that private companies are to serve the high paying subscribers and a few percent tax collected from them will be used to provide service to the masses?

Well, at best one may say that NTP’99 is not a comprehensive policy towards Telecom and IT for all. It has addressed an immediate problem and found the best available means to solve it. One has to go beyond NTP’99 to look at several of these problems afresh.

It is often believed that the key to expansion of network is

• technology and
• finance.

There is a prevailing assumption that both these components are not available in India. The national policy is therefore naturally designed to attract multinational corporations with their technology and finance. This is considered crucial at this stage of infrastructure development for the country.

However, the above assumption has a basic fallacy. One does not have a quarrel with multinationals, but market and affordability in India has unfortunately not been considered a basic starting point; and market and affordability is the key if one wishes to install hundreds of million telephone and Internet connections in India.

Today it costs in India about Rs.32,000 to install a telephone line [Report], [Jhunjhunwala, '98a]. Internet uses the telephone network and access to a telephone is a must, if one has to have Internet. Now, if one takes 15% as the finance cost and 15% as operation, maintenance and obsolescence cost (the operation and obsolescence cost will in fact be higher), it requires an income of 30% of Rs.32,000 per year from each line for the operator to just break even; and this does not include any license fee, interconnectivity charges and spectrum charges. This amounts to Rs.9,600 per year or Rs.800 per month. Less than 2% of the population in India can spend this amount and at the current cost, even with cross-subsidy, not more than 3% could afford telephones. This simple computation stares us interface, but is often ignored.

Will foreign technology help us to overcome this problem? The problem is that the cost of installing a telephone line in the West itself is $1000. Is this likely to drop substantially in coming years? It does not appear so. A $1000 investment requires an income of $300 per year taking into account finance, operation and obsolescence. At $25 per month, it perhaps is affordable to about 95% of the people in the West. Therefore, there is no incentive to reduce cost. Instead to stay in the market, the telecom companies in the West have to attempt to provide a larger basket of services at the same cost. The technology emphasis is therefore rightly to provide more services and features, keeping the cost more and less constant. It is a story similar to what one finds in the PC world. The cost of the PC has stayed constant for the last 15 years inspite of major technological developments; the only difference is that which one could earlier get a bare PC (using 8088 and without a hard disk), and now one can get a full-featured Pentium machine at the same cost.

In India, we require telecom technology affordable at large. Of course, sharing of lines enhances affordability significantly[4]. But to provide telecom service even to 25% of the homes, one need to get telecom technology which would cost, say no more than Rs.10,000 per line. At such an investment, one needs a revenue of only about Rs.250 per line per month to breakeven. With cross-subsidy, the revenue required could go down to Rs.100 per month; and this is affordable to about 20% of our population. We suddenly are talking about 100 to 200 million telephones and Internet connections in India.

It is true that technology for Rs.10,000 per line telephone and Internet connection is not available today. One may question whether such a drastic reduction from around Rs.32,000 per line is really possible. The answer to such a question is simply, why not? Who has attempted this? The R&D community in the West have had a totally different focus. Many of their innovations and inventions may be very useful. It is upto R&D groups with this focus to use whatever technique is available, use their creativity and zeal, and move towards such a goal. The goal may not be achievable immediately, but there is no reason to believe that one can not approach these numbers with time. It will not be easy, but then when have R&D tasks been easy? [Jhunjhunwala, '99]

To corroborate that this is possible, we briefly describe the efforts of the Telecommunication and Computer Network (TeNeT) group in Appendix A. This group at IIT Madras, along with several companies formed by alumini of IITM, has been carrying out R&D to develop Telecom and Computer Networking technologies for developing countries like India. The aim is to significantly reduce cost such that 100 to 200 million telephones and Internet connectivity becomes a reality in India. With about five years (and about 1000 man-years) of effort behind it, the group is currently planning city-wide telecom and Internet network for operators, such that cost per line is close to Rs.18,000. The solution provides simultaneous voice and Internet service for over 50% of connections. The cost is likely to drop to about Rs.15,000 per line in about a year. Beyond this, new technological developments will be required to further reduce costs. What is important is that the costs are not much higher in rural areas. It costs only about Rs.3000 to Rs.4000 more per line to provide voice plus 35/70 kbps Internet in rural areas. [Jhunjhunwala, '98b.]

A note of caution is in order. The efforts of TeNeT group demonstrates that it is possible to significantly reduce cost and create a basis for providing telecom and Internet connectivity to large sections of people. If the dream of 100 to 200 million telephones and Internet connectivity is however to materialise, several similar, yet unique, efforts are required [5]. Fortunately, there are a large number of companies in India who have the technical ability to take up such a task. Most of them today concentrate on providing service to companies in the West and thus work for Western market. They do not believe that this large Indian market can really be tapped—they are afraid that the beauracratic wrangling in the country will prevent such a thing. The success of TeNeT group in large-scale deployment of its technologies is therefore crucial. It will result in confidence-building amongst many in India; the process of the Indian IT and Telecom companies looking inwards and developing products affordable in India (and other developing countries) could easily take off.

To sum up, availability of affordable telecom and Internet solutions is the basic requirement to universal connectivity in India (and most other developing countries). As such products are not avaialble in West, the development of these products became a key priority.

Is availability of these cost-reducing technologies the only bottleneck to fast expansion of Telecom network in the country? The question is whether such technologies would be acceptable in the country and can adequate financial resources required be generated?

Let us look at the switches developed by Centre for Development of Telematics (CDOT). This organisation designed and developed switches in India in late eighties. It faced serious obstacles in its initial years, with the multinational lobby, along with a few officials, running down the product using all kinds of arguments. However, the persistence of those who believed that indigeneous development is the key to affordability, paid in the long run. By the end of the current year, 10 million out of the total of 20 million installed telephone lines in India will use CDOT exchanges. The CDOT exchanges are working in rural areas, small towns and even in large cities. The product has been proven in time to be as reliable as any imported exchange, and in fact have proved to be better in the harsh Indian environment. Today, the exchanges have been exhaustively tested for upto 40,000 line capacity. The exchanges have been upgraded with signalling software for SS7, V5.1, and V5.2, ISDN and even Intelligent Network (IN) services. The most important thing is that today a CDOT line costs Rs.2700 as opposed to Rs.4300 per line for imported exchanges; and in fact if CDOT exchanges were not available, the cost of an imported exchange would have probably exceeded Rs.6000 per line. [6]

However, this CDOT exchange has so far not been used by even one Basic Service Operator. Why?

The answer to the question could easily be clouded in all kinds of technical jargon. The facts however are that 50% of the telecom lines in India uses CDOT exchange today. Any field level DoT engineer will swear to its reliability. The costs are less than two-third of that of imported exchange. The answer therefore lies elsewhere.

The first reason is that many times, the foreign partners of the Basic Services Operators (BSO), would like to sell their own or their partner company’s product. They therefore prevail on the Basic Services operators to buy the more expensive product.

The more important reason is finance. Multinational companies offer vendor financing for their products. In fact they are ready to provide even equity financing, if the BSO are willing to use their products. The finance terms they offer is attractive — LIBOR (International bank rate of around 6%) plus 0.5% interest rate and, more important, a five-year mortorium on all payments. These terms are extremely attractive. Most of the Basic Service Operators are short of finance. A five-year martorium on payment implies payment only after they are likely to break even and are in position to pay. So what if the imported products are more expensive?

But would not the expensive imports make the whole project more expensive? Would not only high-paying customers be then provided the service? Would the country and subscribers not pay in long run? Would not the goals of National Telecom policy be defeated?

There is no doubt that finance is power. Finance gets decision taken. What should the country do in such a situation? When the country loses in the long run due to certain financial imperatives, it has to use policy as a weapon to overcome this disadvantage. More on this in section 4.

Let us deal here with the more general question of whether there are financing options for Indian telecom and Internet expansion.

Some of the most innovative initiatives that have been taken in India over the last ten years, include launching of STD-PCO, Cable TV operation and Computer Training Institutes. These have mushroomed all over the country. The cable TV operation is probably the most widespread. Today cable TV reaches some 35 million homes in India. It was non-existent in 1992. The important thing is that all this happened without any initiatives from Government, any R&D institute, any multinational, any large Indian corporate house and without any support from World Bank or Asian Development Bank. It happened because a couple of youngsters in each locality would bring-in Rs.300,000 to Rs.500,000 by borrowing money and even mortaging family jewellery. They would provide service in a neighbourhood, going house to house to sell connections and going there every month to collect Rs.100 to Rs.150 each. They live in the neighbourhood and are known people. When something goes wrong, one has only to go down the street corner, and complain, or even have a shoutdown. One gets service days and nights, week-days and weekends. If the service is not good enough, one just opts for another cable TV operator; one pays at the end of the month at one’s doorstep only if one is satisfied. This is what Mahatma Gandhi referred to as face-to-face society.[7]

Leaving the implication this has in terms of service aside, let us examine the financing issue. With about 35 million cable TV connections existing and taking average cost of Rs.300 for a connection, the total investment in this operation is of the order of Rs.10 billion (or Rs.1000 crore). Where has this investment come from? Is it not an excellent example of internal resource mobilisation.

This is not to argue that telecom in India does not need international finance. In fact, all financial resources need to be tapped and international finance may be key. But financing must not be allowed to force expensive network choices. Having support and viable alternate sources of local finance will form international finance also to be deployed cost-effectively.

Having discussed some of the key issues which need to be tackled to enable fast expansion of the telecom and Internet network in India, let us get back to NTP’99. What needs to be done now in terms of policies? We take up some specific issues.

It is clear that National Telecom policy must focus on reduction of network cost, a must if universal communication is to be made possible in India in a short time. How is this to be done?

Can incentives be provided to operators for using low cost indigenous products? The issue will be what kinds of incentives and to what extent. The incentive could be in terms of better revenue-sharing terms to the operators for using such products. For example, could the operators using CDOT's exchanges and IITM (TeNeT) group's Access product be given a more favourable revenue-share as compared to those who do not? Can incentives be given to the extent so as to compensate for favourable vendor financing terms available for imported equipment? Alternatively incentives can be provided in terms of lower or deferred entry fee or deferred revenue sharing payments.

One has to look at WTO agreements and other commitments that the country has made while proceeding on this line. But as the incentives are only to compensate for an attractive vendor finance package available for imports, it should be possible to find a way.

If 100 million lines are to be commissioned over the next 8 to 10 years and even if we assume a modest cost of Rs.15,000 per line, the total cost works out to be Rs.1500 billion. Where will all this equipment come from? Will this be the import bill? Going by the current volume of manufacturing in India and the trend adopted by the Basic Services Operators, it appears that all equipment will be imported. Can India afford this?

If not, what kind of incentives be provided for local manufacturing? Can such incentives be built into the license agreement? Once again one needs to study the WTO regulations and find a way. If we do not, we simply will not be able to have 100 million Telephones and Internet connections.

NTP '99 did attempt to make the operations of Cellular and Basic Services Operators viable by converting the committed license fees to revenue sharing. But there is no reason why it has recommended entry fee as a barrier for new operators to come in. Why have such barriers at all? Why not use the same policy as adopted for Internet Service?

Secondly, why should one continue to think in terms of a state-wide (circle-wide) operator only? Why not think in terms of smaller operators.

As discussed in Section 3.3, some of the most successful operations in India have been Cable TV and STD PCO operations. Why not allow similar operations in telecom? Why not allow small operators (franchise operators) providing 500 to 1000 line connections in a neighborhood?

Today, the access network costs 65 to 70% of the total per-line cost of the telecom network. Further, it is the access part of the network, which requires maximum care and maintenance. As this kind of service is best provided by a local person in a neighbourhood, it is best to have a franchise operator put up the access network and provide the service. The backbone networks could be provided by a state-wide operator (DoT or a BSO). As such a franchise operator puts up in 65 to 70% of the cost, he/she must get at least 50% of the revenue.

Why cannot the NTP specifically favour such an arrangement?

There should be specific incentives to encourage an operator to provide service as widely as possible. Some suggestions are:

• incentives to be given to provide service in small towns and rural areas.

• revenue share accruing to the Government for operation in small towns should be zero
• revenue share accruing to the Government for operation in rural areas should be negative, implying that the operator can offset part of the revenue share that it has to pay to the Government for its operation in urban areas by providing more telephones in rural areas.

• to discourage an operator from focusing only on high paying customers in urban areas, the revenue share to be paid should decrease as telephone density provided by the operator in a city increases. In other words, let the operator benefits by providing service to the middle class and lower middle class.

It is important to point out that Universal Service Tax can not compensate for universal service obligations. But incentive to widen customer base would greatly help.

In the international arena, the difference between call minutes originating from one country to another and call minutes terminating in the first country from the other, are used for a settlement. In other words, as many more call-minutes originate from USA to India as compared to the other way around, VSNL actually gets paid by US operators.

The reasons why more calls originate in one country rather than another varies. Of course lower tariffs in one country is often the reason. But the other main reason is that the income level in one country may be higher than the other. For example, when Indians leave India and settle in USA, they earn in dollars and therefore much more in absolute terms as compared to their relatives in India. Even if the tariff in two countries were same (in absolute terms), many more calls are likely to be made from USA to India rather than the other way round. However, it requires a telephone in both countries to complete the call. Recognising this, the international settlement is carried out.

Similar situations exist within India between rural and urban areas. Very often the family
bread – winner leave the village and comes to a town to work. It is he/she who has much more money to make the call to the village.

It is imperative that settlement between the incoming and outgoing call-minutes in rural areas be carried out. Otherwise, the operators who serve rural and back-ward areas of the country will be at disadvantage.

Some policy initiatives to strengthen NTP '99, and render its goals more achievable, have been highlighted. There are several other specific policy decisions that need to be taken as we move on. For example, it is surprising that NTP '99 did not try to resolve the issue of mobile vs fixed wireless service tangle. Still NTP did what was needed. The point is to move on. The main goal of the telecom policy needs a emphasis once again — to provide reliable telephone and Internet connectivity to everyone.

Such goals are achievable only with clear policy initiatives. A push is required such that Indian companies and institutes take up the task to design and develop telecom and IT affordable to people in India. Initiatives need to be taken to encourage large scale efficient manufacturing in India. Finally, policy instruments need to be used to provide incentives to expand telecom and Internet operations in small town, in rural areas and to provide service to lower-income groups in urban areas. Only such explicit policies will enable India to expand its network to the required extent.

In recent years, developments in the area of fibre optics and microwave radio technology, have reduced the cost of the backbone telecom network to about Rs.1,000 – 1,500 per line. Similarly with the access network getting separated from the exchange, the cost of the main exchange today is only around Rs.1,200 per line. The key contributor to the telecom network cost today is the access network, which is sometimes as high as Rs.22,000 in some urban areas. The cost in rural areas may be much higher.

The TeNeT Group and its associates which include Midas Communication Technologies, , Banyan Networks, Vembu Systems and Nilgiri Networks, have developed a Fibre Access Network (optiMA), Wireless in Local Loop system (corDECT) and a Direct Internet Access System (DIAS), which aim to significantly reduce access cost and at the same time enable large scale usage of Internet. Using these technologies it is possible to set up a total network today at an average cost of Rs.13,500 per line. These systems flexibly enable rapid expansion of the telecom and Internet network both in urban and rural areas. Let us first take a brief look at these system technologies.

A.1 optiMA Fibre Access Network

This fibre-to-the-curb (or street-corner) system provides one of the most cost-effective means of deploying telephones especially in dense urban areas. The system consists of a Remote Terminal (RT) deployed at street-corners as shown in Fig.A.1. The single cabinet with a built in power plant and battery, serves about 500 subscribers and could be located at a PCO or a street-corner kiosk. The subscribers are connected to the RT on copper (either POTS or ISDN) or on wireless (the last 500-800 m). The RTs on the street-corner are connected to a Central Office Mux (COMUX), located at the main exchange premises using a fibre-optic ring as shown in Fig.A.2. The ring network enables the system to withstand any single failure of fibre link. The COMUX is connected to the main exchange on standard E1 lines using V5.2 protocol and to the radio exchange (DIU) to serve wireless subscribers. As shown in Fig.A.2, a Remote Access Switch with Modems (RASM) connected to the main exchange on E1 would ensure that Internet traffic from different users are statistically multiplexed before entering the backbone telecom network. The RASM also enables a guaranteed 56 kbps Internet connectivity as the analog portion of the loop is now reduced to 500 - 800m. The cost of the POTS access solution using optiMA is astoundingly low, approximately Rs.7,000 per line including Rs.5000 for copper cost in the last 800 m. The wireless connection would cost around Rs.13,000 and the use of multiline unit providing connections to four subscribers in a building would bring down the cost to around Rs.8,000 per line.

corDECT Wireless in Local Loop system, based on the DECT standard, has an interesting architecture, especially for its fixed part. The fixed part consists of a DECT Interface Unit (DIU) acting as a 1000 – line wireless switching unit providing a V5.2 interface towards the main exchange, and weather-proof Compact Base Stations (CBSs). These are connected to the DIU on either three pairs of copper wire carrying signal as well as power, or fibre/radio using E1 links through a base station distributor (BSD). The subscriber terminal is a wallset (WS), with either a built-in antenna or a rooftop antenna providing a line-of-sight link to a CBS. The WS has an interface for a standard telephone (or fax machine, modem or payphone) and an additional RS232 interface for a computer, enabling Internet connection at 35 or 70 kbps. No modem is required for Internet access, since all links between the WS and DIU are digital, as shown in Fig A.3.

Efficient transmission of packet-switched Internet data on a circuit-switched network is achieved by combining a RAS with the corDECT system. The connection is digital all the way from subscriber to ISP. The Internet call from a WS to a RAS does not enter the exchange at all, but terminates in the access network itself. Only the concentrated IP traffic from the RAS to the ISP traverses through the exchange and PSTN.

All subsystems are built primarily using digital signal processors (DSPs), with the DIU having nearly 100 DSPs. This soft solution, while cutting down development time and providing design flexibility, also ensures that the cost of the fixed part is no more than 15% of the total per-line cost in a fully loaded corDECT system. This allows deployment flexibility for both dense urban and sparse rural areas.

A new operator who wishes to initially deploy lines in a mid-sized town or city in the very first year would use the deployment scenario using a cluster of about 12-15 CBSs (each serving 50-70 subscribers at 0.1 Erlangs each), along with the backhaul microwave equipment, mounted on a 15m rooftop tower to serve an area of 2-3 km. This deployment uses no cables and can be made operational in two to three months at a total deployed cost of Rs.15,000 per subscriber.

Later, the operator could increase the number of lines by using an optical fibre grid to connect BSDs to the DIUs. A CBS cluster now serves 1,000 subscribers within a 700m to 1 km radius. Here, many subscriber installations may not need line-of-sight links to the CBS. Once again, the total deployed cost of the access solution is under Rs.15,000 per subscriber, including the cost of optical fibre cable and cable-laying.

The corDECT system also offers an excellent deployment opportunity for a small town and its surrounding rural areas at a similar cost. The mode of deployment uses DIU at the tower base. To serve about 1,000 subscribers, an operator needs a tower (about 35m high) in the town centre. The microwave link connects the DIU to the nearest trunk exchange. The base stations now serve subscribers within a radius of 10 km using wallsets with rooftop antennas providing line-of-sight links.

Deployment in sparse rural areas is possible using the corDECT Relay Base Station (RBS). A two-hop DECT link is used to provide connection to the subscriber as shown in Fig.A.4. One link is from the WS to the RBS, which is mounted on a tower typically 25m in height. The other DECT link is from the RBS to CBS, which is also mounted on a tall tower (say 40m). Both the RBS and CBS use high-gain directional antennas, making a 25km link possible. The 5km maximum link distance due to the guard-time limitation of DECT is overcome by the use of auto-ranging and timing adjustment. This technique is used in the RBS to support a 25km link, and to enhance the CBS range to 10km. This provides a subscriber density as low as 0.5 subscriber/km² at a total cost of Rs.18,000 per line.

The Direct Internet Access System (DIAS) allows service providers to provide high bandwidth Internet access to residential and corporate subscribers, in addition to voice services, without any changes to the existing cabling infrastructure. In contrast to current residential PSTN (Public Switched Telephone Network) and ISDN (Integrated Switched Digital Network) dial-up access, the DIAS provides an Always On Internet Access that is permanently available at the customer's premises. Using DSL techniques, seamless voice and data connectivity is provided to the customer over the same pair of copper wires.

Implementation of this system is done using the existing cable plant. All that is required is the installation of the IAN (Integrated Access Node) at the exchange and a DSU (Digital Subscriber Unit) at the customer premises as shown in Fig A.5.

The DIAS has a DSU that combines voice and data packets on a single twisted-pair wire at the subscriber's premises. At the service provider's premises, an IAN separates the voice and data traffic from a number of subscribers and routes them independently to the PSTN and the Internet respectively. The IAN is connected to the PSTN via the E1 V5.2 port, and to the Internet either through 2 E1 data ports or through the Ethernet port. Alternatively, the PSTN connectivity can be achieved through the POTS ports of the exchange with the addition of an optional subscriber multiplexer module, that converts a single E1 line into multiple POTS lines.

The DIAS system provides 2 types of voice and data services:

• The BDSU (Basic Digital Subscriber Unit) is designed for the SOHO (Small Office Home Office) and residential Internet user. It provides a permanent Internet connection at a maximum data rate of 144 Kbps, which drops to 80 Kbps when the telephone is in use and transparently goes back to 144 Kbps when the telephone is not in use.
• The HDSU (High bitrate Digital Subscriber Unit), which is designed for corporate subscribers, can provide voice connectivity for upto 8 telephones and a permanent data bandwidth of upto 2 Mbps, which drops by 64 Kbps for each telephone call.

Before we proceed to describe some of the deployment scenarios using these technologies let us look at a problem associated with accessing the Internet. The most common way of accessing Internet today from home and office uses the telephone network. One just has to connect a telephone modem and a computer to a telephone line and dial to a modem bank at Internet Service Provider (ISP) as shown in Fig.A.6. The modem bank connects the user's computer to the ISP Router, which is connected to other routers on the Internet. This simple-looking scheme has a few problems.

Telephone calls are usually for a few minutes each and the average traffic from a telephone is about 0.1 Erlang. The telephone network in most places, but specially in developing countries, is designed to handle just this much traffic. Now, the Internet calls are usually of much longer duration, typically lasting 30 to 60 minutes. The traffic that Internet calls inject on PSTN, especially during busy hours, is much more than 0.1 Erlang per Internet subscriber. Fortunately only a very small fraction of telephone users in developing countries use Internet today. As more and more people get on the Internet, increasing traffic beyond 0.1E/subscriber, the network would get congested and collapse.

The solution to this Internet tangle lies in recognising that even though Internet usage is for long hours, the traffic is bursty. With long silence periods between traffic bursts, rarely the average traffic on Internet exceeds more than 10 to 15% of the peak traffic. If one could only statistically multiplex Internet traffic from multiple subscribers, the total traffic from Internet (with peak access rate less than 64 kbps) is unlikely to significantly exceed the voice traffic. The access network framework that therefore emerges is shown in Fig.A.7.

An Access Centre (AC) placed in an area provide services to the subscribers in nearby vicinity. The subscribers premises have a Subscriber Unit (SU) providing a standard telephone interface (RJ-11) and a port (either serial port or Ethernet port) to connect a PC. It is assumed that the access at subscriber's premises would be usually a digital access, providing direct digital connectivity to the PC (without an external modem). The interconnection between AC and SU could use any of the multiple mediums - copper, DSL on copper, [Cioffic], digital wireless, fibre [Burpee] coaxial cable or even power-line [Perkins]. The Access Centre could in fact combine the line interfaces for these various mediums. The key task at the AC is to separate the voice traffic and the Internet traffic. The voice traffic is to be sent to an exchange usually using one or more standard E1 interfaces with V5.2 signalling. The Internet traffic is statistically multiplexed and transmitted to an ISP on usually one or more E1s.

How large an area and how many subscribers should an AC serve? Ideally one would like the area to be small, as it is the access from AC to SU that may dominate cost. Also, small distances, especially on twisted pair copper, implies that higher bit-rate DSL can be used and upgradations will be possible in future. However, installation of AC requires some infrastructure (especially power back-up) and the cost of this infrastructure must be taken into account.

Access Centres need to be connected to the Central Switching Center where main exchange and routers are located. This interface is best provided by a fibre backhaul. As multiple ACs are to be located in an area, it would be desirable that a fibre loop connects ACs to the Central Switching Centre. However, in order to get quickly started or for interfacing a AC in rural area, an operator may prefer digital microwave radio. Cost of such PDH as well as SDH radios have come down by a factor of about 5 in the last five years, due to developments in the area of MMICs and Digital Signal Processors.

In urban areas, fibre backhaul is the obvious choice. A PDH or SDH fibre ring network with Drop and Insert as shown in Fig.A.8 could be used. At each Drop and Insert, multiple E1s are dropped to connect an AC to both the main exchange as well as to the Router. The dual ring architecture provides the redundancy and a single point failure does not bring down an Access Centre. The optiMA Fibre in the Loop system is an ideal choice to provide this connectivity.

A.5. The Last Mile

As discussed earlier, there are multiple options for the so called last mile - the interconnection between the AC and the subscriber unit. If POTS service is all that is required, and the distance between AC and SU is less than 800m, the most inexpensive access solution is provided by plain old copper. Wireless access is the next best solution. The DSL on copper provides high quality service at slightly extra cost. A DSL solution that is becoming popular is the one using Basic Rate DSL (BDSL) which provides 144/80 kbps Internet connectivity plus telephone. Of course, HDSL and ADSL provides access at much higher bit-rates.

The co-axial feeder could provide an excellent Access Solution providing telephone, Internet as well as video. However, the experiments undertaken in India have not been very encouraging. The problem is due to poor quality of cables and connectors used, higher man-made noise in 5 MHz to 45 MHz frequency band and due to illegal taps which significantly adds to the noise in the uplink direction. Further experiments are underway. The fibre in the last mile is too expensive currently for most developing countries.

The solution now being deployed widely uses an Access Centre as shown in Fig.A.9 using Access products developed by TeNeT group. The access centre provides POTS, wireless and DSL service. The wireless subsystem, referred to as corDECT, provide 35 kbps Internet plus simultaneous voice connectivity on WS-IP. A Multi-Wallset (MWS) providing telephones to four subscribers in a multistorey unit bringing down the cost significantly. The DSL subsystem, referred to as DIAS, provides 144/80 kbps Internet connectivity plus voice using BDSU and 2 Mbps Internet connectivity plus 8 telephones using HDSU. In addition, POTS connectivity is provided at low cost.

A complete access network providing 1100 WS-IP subscribers, 400 MWS subscribers, 180 BDSU subscribers and 480 POTS subscribers at each Access Centre costs around Rs.13,500 per line. The cost includes about 4 km of fibre associated with each AC, SDH Comux and Drop and Insert as well as the cost of fibre laying, copper laying and installation cost. Further, the cost includes about 25% taxes.

A.6. Deployment Examples

The deployment of Access Centres varies a lot depending upon the area that need to be served. Fig.A.10 shows a plan of providing 16 Access Centres in city of Chennai in India. Each Access Centre would typically serve a area of 1.6 km by 1.6 km, but wireless connection can be given to a much larger area when required. The SDH fibre connects the Access Centres to a central switching unit. Each Access Centre will serve about 2000 subscribers. Fig.A.11 shows deployment in a rural area, a district of Thanjavur, a state in South India. Here, the backhaul links could be either by microwave radio or a fibre. The objective here is to provide at least 35 kbps Internet connectivity in addition to voice at every village. A Relay Base Station is used to extend the reach of wireless to 30 km. The cost of serving about 15,000 subscribers spread over the whole district is around Rs.20,000 per line. The network can be set up in about a year.

1. The day after the announcement of new rates by TRAI, there was a hue and cry in the Parliament, with some of the parties supporting the ruling coalition threatening to walk out. The Government withdrew the revised tariff which was brought back after some revision a few weeks later.
2. Almost all cellular operators in the country, except those operating in Delhi and Mumbai, have a monthly income from each subscriber far less than what was envisaged. There are many people who keep cellular phone, but use it only sparingly. Even the upper middle class finds the cellular air-time charges too high.
3. Unfortunately however, all private operators are aiming at exactly this, particularly in the first few years of operation. One cannot fault them as it is a source business policy.
4. An example of such sharing is the STD PCOs, which is available today at all nooks and corners of at least urban India. Around 15% of population finds this affordable, and even use long-distance calls at late nights. It generates around 20% of total telecom revenue.
5. For example, bringing Internet connection to home still requires massive investment in a computer, which well over 95% of families in India cannot afford. Low cost Access terminals using TV or an integrated telephone, TV and Internet Access terminal definitely need to be developed
6. For a year or two in mid nineties, CDOT exchange manufacturers were in financial trouble an DoT was not buying CDOT exchanges, the multinationals had suddenly significantly increased the per line price of their exchanges.
7. This is a running theme in several works of Mahatma Gandhi. He contrasted this face-to-face service with the service provided by a beauracratic faceless body, to which you can only petition for a service. The methods advocated by Gandhiji definitely needs a closer look, specifically in face of a situation where service from most organised sector in India is either becoming too expensive or is totally breaking down.