Today, many in India believe that Telecom and Internet Access can make a major difference to the nation. There is a strong desire to get better access and people are in general not satisfied with the level of service. Department of Telecommunications (DoT), so far the monopoly operator in the country, is trying to expand its network rapidly, adding almost 5 million telephones in the last two years to the 12 million figure in 1996 (see Table 1); but this is still considered inadequate. The 17.5 million telephones for its 970 million people, account for a teledensity of barely 1.5, whereas the world average is about 14 and even many of the developing countries have a teledensity of more than 4. About 25% of the telephones are in four metros. Eventhough 74% of the population lives in rural areas, the number of village telephones is less than 1% of the total. The number of Internet subscribers amount to barely 1 per 10,000 population and mostly confined to large cities. It is in this situation that a few years back, the Government of India decided to bring in a private Basic Services Operator (BSO) to compete with DoT. Expected to bring in required finance, one BSO was licensed in each state. There was a general belief that the wait for telecom and Internet access will be a thing of the past.

The only way the middle class and the people in rural areas [2] can be provided telephones and Internet connection in India, without large-scale subsidy, is by reducing the telecom infrastructure cost. If the per-line cost can be brought down somehow to Rs.15,000 (from Rs.30,000 today), rapid expansion of the telecom network can take place. The question is: can technology be used in an innovative manner to make this possible?

This paper examines this question with an aim of achieving hundred million telephones and twenty-five million Internet subscribers in India in a short span. It points out that, now the BSOs are starting afresh, they need not follow the conventional path. A number of Technological developments have taken place in the recent years, which can be taken advantage of. What is however required is harnessing of these technological developments to make and deploy affordable products. That these products have not been deployed by a operator in the West is understandable, as the West has to live with the legacy of already having the telecom infrastructure and its only choice is build upon that DoT, to some extent, suffers from similar disadvantage, but can definitely use these technologies for further expansion.

Section 2 of this paper deals with the new paradigm of telecom network. The focus is on new access Technologies and overlay of efficient Internet service on the network, that breaks new ground in improving efficiency and reduces cost. In section 3, we look at Indian attempts to harness technologies and standards, evolved often for different purposes, towards achieving low cost and richness of features. Section 4 concludes the article.

We now examine the impact of changing technology on the major components of the telephone network, namely, the backbone, the switches, and the access network.

Until the eighties, the Indian long distance telephone network mostly used 4-wire or 6-wire analog trunks, along with FDM analog multiplexers. The total number of circuits available were barely 50,000 [3], and the quality of service provided by these links was considered unsatisfactory. One often had to wait for hours for a long distance trunk call to mature and then often shout to be heard on the other side compensating for the poor signal to Noise Ratio on the long distance lines. The introduction of digital multiplexing and later of optical fibres changed this situation. While the quality dramatically improved, the cost to the operator also continuously went down.

Today, there are a number of manufacturers of optical fibre in the country. The cost of a 12-core burial-type optical fibre cable has come down from around Rs.200,000 per km a few years ago, to barely Rs.40,000 per km. SDH-1, SDH-4 and SDH-16 multiplexers and drop-and-insert equipment are being rapidly introduced in the network. WDM is however not being used in a significant way yet, as the emphasis today is to lay new fibre rather than expand capacity on existing fibre. It is recognised that WDM can be introduced in future to multiply capacity.

Besides optical fibre, point-to-point digital microwave radio systems have played a significant role in expanding the Indian backbone network, and continue to do so. Useful where fast deployment is required or where the terrain is difficult, or to serve low-density or medium-density rural areas, these links are deployed with 8 Mbps, 34 Mbps, 140 Mbps, and now SDH-1, capacity. Several frequencies are used in the 2 GHz, 7 GHz, 11 GHz, 13 GHz, 15 GHz and 18 GHz band. Point-to-point radios are especially popular for short-haul feeder links. Much of the equipment is manufactured in India. The costs have again tumbled down over the years, with a 34 Mbps 2 GHz 30 km link costing about Rs.1.5 million today including installation and tower costs.

To sum up, the backbone technology is mature and the emphasis today, rightly enough, is on the induction of a Q3-based Network Management System [4] for the network, even as it grows. The cost for providing a new high-capacity backbone network works out to barely Rs.1,500 per subscriber line, assuming conventional traffic levels.

Up until the early eighties, most of the exchanges in the Indian Telecom Network were electro-mechanical exchanges. The exchanges were expensive and the service could not be expanded to meet the demand primarily because of limited capital. As a result, most the exchanges were overloaded. Also, as very few telephones were available, there was significant usage of lines on a shared basis. The traffic generated was higher than the 0.1 Erlang per subscriber for which these exchanges were designed. Call blocking therefore increased significantly during the peak hour, providing poor quality of service.

It is here that the innovation of time-switching [5] came to the nation’s rescue. Digital exchanges, that are essentially non-blocking, have since then been introduced rapidly. The size, power consumption, and the cost-per-line of the exchange, have come crashing down.

The digital electronic exchange today consists of the following : switching matrix, processors (computers) with associated software, and subscriber and trunk line interfaces. The revolution in IC technology has made the switching matrix of even large exchanges simple and inexpensive. The line interface used on the trunk side today is the digital multiplexed 2.048 Mbps E1 interface (ITU-T G703). Single-chip solutions for this interface have made its cost low. The introduction of Remote Line Units, as discussed in Section 2.3, has led to a division of the exchange into a Switch Core (also referred to as Main Exchange) and the Access Unit (AU). The subscriber interface in the main exchange, also, is the low cost digital multiplexed E1 interface to the AU. The key of the Switch Core (main exchange) technology is therefore the processor and the software. Today, it is imperative that a switch core has Signalling System 7 (SS7) software, ISDN signalling software, advanced Operations and Management (including Network Management) software, and the ability to handle a fairly large number (say, 20 per line) of Busy Hour Call Attempts (BHCA). Further, it is also important to have provision for Intelligent Network (IN) services. But most important, the main exchange should have V5.1 and V5.2 [6] protocol software to enable a third-party AU to be connected seamlessly to the exchange.

With the Main Exchange technology becoming primarily software-intensive, the cost per line comes down when the usage base is high. It is expected that the Main Exchange without the Access Unit will cost less than Rs.1,000 per line in the near future.

Till the eighties, an exchange in an Indian urban area used to typically serve subscribers located upto 8 kms away. Consequently, the twisted pair copper used in the local loop would typically be 0.51 mm (SWG 24) gauge or higher. It was difficult to maintain such long loops and as the copper prices rose every year (due to inflation), the cost of the local loop became a significant part of the total cost of the telephone network. The development of Remote Line Unit (RLU) and Remote Switching Unit (RSU), therefore, came as a big boon. These units, connected to the main exchange using digital multiplexed E1 lines, often using optical fibre, would now serve subscribers located at most 3–4 kms away. Not only is the local loop shorter, significantly reducing the cost, but the RLUs also concentrate traffic, something hitherto done only at the exchange. If a subscriber generates 0.1 Erlang traffic, a 1000–subscriber RLU would have a total traffic of 100 Erlangs and would only require 4 E1 links (120 lines) to the main exchange to provide 0.5% grade of service. This not only reduces the link capacity required between the main exchange and the RLU, but also reduces the size of the switching matrix required at the main exchange.

The RLUs and RSUs, introduced from the late eighties onwards in India, reduced the cost of access to almost two-thirds of its earlier level and contributed to keeping the per-line cost at a near constant level, despite inflation. However, there was one problem which did not allow one to take a full advantage of this innovative concept. Although the RLU-exchange link was provided by a standard E1 interface, the signalling protocol used (hereafter referred to as the Access Protocol) was defined by the manufacturer and remained proprietary. With the subscriber line interface removed from the main exchange to the RLU, and with the switching matrix of the main exchange reducing in size (due to concentration at the RLU), most of the cost shifted from the main exchange to the RLU. Of course, as discussed in Section 3.2, the intelligence, in the form of a large part of the software, remained at the main exchange. If the access protocol was standardised, one could marry new, third-party, low-cost RLU technology with the main exchange.

The switch manufacturers, fearing loss of revenue, preferred that the access protocol be proprietary, even though there was little technical justification for this. Fortunately, ETSI, followed by ITU-T, took initiatives to standardise these protocols, referred to as V5.1 and V5.2 access protocols [6]. The V5.1 protocol is used when no concentration is carried out at the AU, whereas V5.2 is used when concentration is carried out. For countries like India, it was again a boon, as they could now source/develop the most suitable access technologies, while obtaining the very best Main Exchange from the switch manufacturers.

Even though reduction of the local loop distance by using the RLU brought down the access cost, by the mid-nineties the rising cost of copper caught up once again. Even the 3 km copper loop became expensive. Today, the total access cost, including the cost of fibre to connect the exchange to the RLU, the cost of the RLU, and that of the copper local loop (including cable-laying charges), is typically Rs.14,000 to Rs.20,000 per line. Furthermore, the cost is increasing every year.

Fig. 1: Access using DLC

At the same time, the cost of fibre was continuously going down. The Digital Loop Carrier System, shown in Fig.1, therefore became a cost-effective option. A typical DLC has a Central Office (CO) unit co-located with the exchange, interfacing typically to 500 subscriber lines. The CO unit digitises the signals on the subscriber interface, multiplexes them and transmits it on an optical fibre, which will carry it to a Remote Terminal (RT). These RTs are small units meant to serve about 500 subscribers within a radius of 500-800 m. They are battery backed-up and are designed to be installed on the wayside.

The DLCs, by reducing the copper loop to 500-800 m, are meant to cost lower when compared to the RLU approach. The cost advantage is expected to increase over the years. In fact, the Indian government has contractually insisted that the private BSO use copper only for the last 500 m, in an attempt to ensure that the latest, most cost-effective, and least cumbersome access technologies are deployed.

The DLC described in the previous section is a retrofit to reduce copper in the loop. It does not take advantage of the concepts of the Remote Line Unit and the standardisation of the Access Protocol. In contrast, the Fibre Access Network (FAN) combines the advantage of DLCs and RLUs. As shown in Fig.2, the FAN (also sometimes referred to as Fibre in the Local Loop or FiLL) also has a CO unit, akin to the DLC. However, the interface between the main exchange and the CO unit of the FAN is now an E1 link with V5.1 or V5.2 protocol, and the CO unit is essentially a digital multiplexer. The subscriber line interfaces, both at the exchange as well as the CO unit in Fig.1, serving no useful purpose, are now avoided. It should be pointed out that use of V5.2, rather than V5.1, provides a cost advantage, since traffic concentration is now carried out nearer to the subscribers at the RT.

FAN can be deployed to service multiple RTs in a star or a loop configuration as shown in Fig.2. The loop configuration has an advantage, since failure of an RT or a cable-cut can be healed by using the reverse link. As in a DLC, the RTs are battery-backed up units, located on the wayside and serving typically 500 subscribers within a radius of 500-800 m.

So far we have discussed the use of copper and fibre in the access network. While a suitable mix of fibre and copper gives a cost-effective solution, a wireless access network is advantageous in many situations. Firstly, wireless access promises quick deployment, an aspect particularly important to the new operator. Secondly, and perhaps more surprisingly, wireless access proves to be more cost-effective then other access technologies as a permanent access network in mid-sized Indian towns and rural areas.

It is important to point out that Wireless in Local Loop (WiLL) should not be confused with mobile communications. Mobile communication systems are designed to provide communication for people on the move, and not in homes and offices. Therefore, mobile systems do not have to cater to the exacting demands placed on the home and office telephone. In mobile phones, voice quality has to be good, but not necessarily toll quality. Data communication is not as important; definitely a high data rate need not be insisted upon. The total number of subscribers to be served is much smaller than that served by the Public Switched Telephone Network (PSTN). The traffic is much lower than in the PSTN. More important, people are ready to pay extra in the form of air-time charges for service while on the move. Of course, mobile telephones have to handle more complex requirements such as universal coverage, hand-offs, and roaming, etc.

The requirements of Wireless in Local Loop are significantly different [7]. Firstly, the quality of service cannot be compromised That this has begun to be understood is evident from the following statement, made recently by an industry representative:

Not only must the voice be toll quality, it is also very important to provide high bit-rate Internet connectivity. While 28.8 kbps data communication requirement is the very minimum, 64 kbps communication is highly desirable. Further, services like N-ISDN are also desirable. Besides, the WiLL system should cater to high traffic levels (0.1 Erlang, or even 0.15 Erlang, per subscriber in view of data communications) and should be able to serve a high density of subscribers in urban areas (as high as 5,000 subscribers per sq. km in several areas). Also, it is desirable that the system serve even sparse rural areas where subscriber density is less than 1 subscriber per sq. km.

While doing all this, the cost must be low. The target installed cost for wireless access in dense urban areas is Rs.12,000 per subscriber and about Rs.18,000 per subscriber for sparse rural areas. If a multisubscriber unit is used, the cost target is as low as Rs.6,000 per line. Note that this cost includes the cost of the Wireless Switching Unit (WSU), Base station, Subscriber unit with POTS as well as data interface and the cost of the backhaul to connect the Base Station (BS) to WSU. It is also important that suitable back-up power is included in the subscriber unit to provide POTS service exceeding 16 hours, because long blackouts are quite common even in large towns.

Today, Internet access is becoming increasingly important. Those who have Internet access have rapid access to all kinds of information, and this could create another divide between the haves and have-nots [2]. A telecom network installed today must provide widespread access to Internet.

Internet access on the existing telephone network appears to be very simple—just connect a modem to your telephone line, dial up a router of an Internet Service Provider (ISP), and get going. Unfortunately, there are several problems in using Internet in this manner, accentuated by the specificities of the telecom network in India. Let us examine these problems.

The PSTN, in many parts of the world and especially in India, has been designed to handle 0.1 Erlang traffic per subscriber. While this is largely sufficient for voice telephony, Internet access complicates the matter. While a voice call lasts only for a few minutes, an Internet call usually lasts much longer [9]. Most studies have shown that an Internet user offers a load as high as 0.3 Erlang during peak hour. As the ratio of the Internet users to the total users grows, the PSTN will just not be able to handle the load. The network will get congested and fail to complete a large number of calls.

The second problem in accessing the Internet by making a switched telephone call to the ISP, has to do with the analogue modem connection between the subscriber and the ISP. The analogue link, in India, is just not reliable, mainly due to the variable quality of the copper local loop. This is even more so when a subscriber is located in a small town, where the trunk could also be analogue. The quality of the dial-up link varies, and while it does provide 28.8 kbps connectivity occasionally, it often provides only 9.6 kbps or 4.8 kbps. Sometimes the modem link also drops, requiring redialling and a new connection.

In India, the solution to this Internet tangle is emerging in the form of a low-cost Remote Access Switch (RAS). Here, one explicitly recognises that the telecom network is a circuit-switched network whereas the Internet is packet-switched. Circuit-switched voice connections occupy a full circuit, but only for a short duration. Internet connections, however, last for much longer durations, but utilisation is very bursty. When such a connection is made over the circuit-switched PSTN, the advantage of bursty traffic cannot be exploited. Yet, the circuit-switched telephone network is the only available access at homes and offices. The Internet cannot avoid this network, especially when millions of connections are to be made.

The RAS solution is similar in principle to Remote Access Vehicles presented by Sain Morgan [9]. RAS equipment is co-located with the local exchange (or even RLUs or RTs of a Fibre Access Network, as we will show later) and connected to it using standard E1 interfaces. A subscriber desiring an Internet connection, dials up the RAS and sets up a circuit-switched local call, as shown in Fig.3. The call uses only the local exchange port of the PSTN. These exchanges today have very little blocking and can therefore handle the much longer holding-times (and therefore, higher Erlang traffic) of an Internet call. When several subscribers set up Internet calls to the RAS, the RAS multiplexes the bursty data from all these subscribers, and routes the data to the ISP using one or more 64 kbps channels. The 64 kbps connections between the RAS and ISP router could be leased or on dial-up basis, and as shown in Fig.3, the calls take the RAS-Exchange-PSTN-ISP route. In fact, it is desirable that the RAS sets up and tears down the connections depending on the traffic volume it encounters.

The call charges can now be low as only intra-exchange calls are being made for such access. The number of connections between the RAS and ISP are now small, and utilise only the reliable digital trunks. No modems are required at either end. Further, multiple subscribers are now being served on each 64 kbps link between a RAS and ISP. Assuming that upto 10 Internet connections use one 64 kbps slot, a single E1 link (consisting of thirty 64 kbps slots) to router could serve 300 Internet calls. The RAS, while providing an attractive solution to Internet tangle, contributes to a very low per-line cost (see Section 3.3).

The key to the smooth functioning of a network with diverse intelligent modules is an advanced Network Management System (NMS). Over the last few years, significant developments have taken place in both computer network management and telecom network management. Today, while the former is largely SNMP-based, the latter is Q3-based [4]. Since the telecom network planned has to cater to both telephony as well as Internet, an integrated NMS is essential. Similarly, advanced Customer Care and Billing Systems are required for integrating billing for telephony and Internet access. As the number of subscribers on the network grows, the impact on the network cost of these software systems would be marginal.

The Indian DoT set up the Centre for Development of Telematics (C-DOT) in the mid-eighties with two immediate goals: one, to develop a small rugged switch which would work in a tropical, non-air-conditioned, environment in small towns and rural areas, and two, to develop a modern large size (40,000 lines) switch with services comparable to those offered by advanced switches.

The first task was achieved very successfully, and most of the deployment in small towns and rural areas in India has been C-DOT switches. Today, there are few comparable products available in the world in the cost range (less than Rs.1,500 per line) of C-DOT's rural exchanges.

The development of a modern large exchange was in line with similar efforts elsewhere. The goal here was not to achieve anything different, but to acquire design capability and reduce cost, capitalising on the software development base in India. Today, C-DOT has a Main Exchange with SS7 signalling, ISDN capability, IN services and a fairly high BHCA. A benefit of the indigenous effort is the ability to react quickly to changing requirements. This long-term vision has already paid-off: quick incorporation of the V5.2 access protocol in the C-DOT switch. When switch manufacturers all over the world are delaying induction of V5.2 in their switches in view of the perceived loss of revenue, C-DOT today has the interface ready and tested for deployment. The C-DOT Main Exchange can support AUs serving upto 50,000 subscribers using the V5.2 access protocol. The cost of C-DOT’s Main Exchange with V5.2 access is close to the target of Rs.1,200 per line mentioned in Section 2.2.

Similarly, there have been several efforts in India over the last five years towards development of digital microwave point-to-point and point-to-multipoint radio systems. Capitalising on the development of microwave monolithic ICs, these efforts have brought down the costs to the low value mentioned in Section 2.1. Similarly, a number of companies have indigenously developed Optical Line Terminating Equipment and Digital Multiplexers for the backbone.

In the mid-nineties, as the cost of the backbone network and switch core reduced substantially, the emphasis shifted to access technologies. Wireless access was and continues to be the most talked about. However, the key to the successful large-scale deployment of Wireless in Local Loop System in India is the right choice of technology. It will not do to have a system which offers quick deployment, but at a high cost. For reasons discussed in Section 2, it is important that the wireless solution chosen has a final deployed cost comparable to and preferably even lower than that of the wired solution. Yet, as pointed out in Section 2.3.4, wireline voice quality and data communications at upwards of 28.8 kbps are required. Further, the system must support subscriber density as high as 5,000 per sq. km. A study of available international wireless standards reveals that the choice narrows down to PCS standards such as DECT, PHS and PACS [7]. These standards can be implemented at low cost, and provide wireline quality, high subscriber density, and high data rate, but have small radio range. While microcellular solutions based on these standards are suitable for dense urban areas, one needs to find innovative deployment strategies in other cases, so as to cover a wide area.

The Telecommunications and Computer Networking (TeNeT) Group at the Indian Institute of Technology Madras (IITM), located at Chennai, has been playing a key role in defining and developing access technologies suitable for India. Along with Midas Communication Technologies (Pvt.) Ltd., Chennai, and in partnership with Analog Devices, USA, for IC development, IITM took up the development of a DECT-based Wireless in Local Loop system.

The system, referred to as corDECT [10], 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. It also consists of weather-proof Compact Base Stations (CBS) connected to the DIU either on three pairs of copper wire carrying signal as well as power, or on fibre/radio using E1 links through a Base Station Distributor (BSD). The DIU, CBS and BSD are built primarily using Digital Signal Processors (DSP), with the DIU having nearly 100 DSP ICs. This soft solution, while cutting down the development time, 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, in turn, allows deployment flexibility, and cost-effective solutions can be found for dense urban areas as well as sparse rural areas.

For example, a new operator who wishes to initially deploy 5000 lines in a mid-sized town/city in the very first year, would use the deployment scenario as shown in Fig.4. All the DIUs are co-located with the main exchange and connected to it using the V5.2/E1 interface. Each DIU is connected to a BSD located on a roof-top at a suitable part of the town using a point-to-point 8 Mbps microwave link. At the BSD site, about 12 - 15 CBS (each serving 50–70 subscribers at 0.1 Erlangs each), along with the micro-wave equipment, are mounted on a 15m roof-top tower to serve an area of 2–3 kms. The subscriber terminal is a wallset (WS), with either a built-in antenna, or a roof-top 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 RS232/V.35 interface for a computer, enabling Internet connection at 28.8/64 kbps. No modem is required as both the wireless link from the WS to the CBS, and the link from the CBS to the DIU, are digital. Digital data is thus routed all the way from the WS to the ISP.

This deployment scenario of 5000 lines uses no cables and can be made operational in 2 to 3 months. What is particularly attractive is that the total deployed cost of the corDECT Wireless Access Network works out to Rs.14,000 per subscriber. Even if the system is not fully loaded to begin with, the cost per line does not increase significantly, since the cost of the fixed part is a small percentage of the total cost.

In later years, the operator can choose to increase the number of lines to a much larger number, using an optical fibre grid for connecting DIUs to BSDs. The total deployed per-line cost does not alter significantly.

The corDECT system also offers an excellent deployment opportunity for a small town and its surrounding rural areas. To serve about 1000 subscribers in a small town, an operator needs a tower (about 35m high), somewhere in the town centre. The DIU is located at the tower-base and the base-stations are on the tower. The DIU is connected to the main exchange located upto 30 kms away using a 8 Mbps microwave link (typically in the 2 GHz frequency band). The base-stations now serve subscribers within a radius of 10 kms using wallsets with roof-top antenna providing line-of-sight links, as shown in Fig.5. The subscriber density served could be as low as 3 subscribers per sq. km, and once again the total deployment cost of the access solution works out to Rs.14,000 per line. Internet connectivity at 28.8/64 kbps can be provided to each subscriber at no additional cost.

Deployment in sparser rural areas is possible using the corDECT Relay Base Station (RBS). The solution provides deployment with subscriber density as low as 0.5 subscriber per sq. km [2] at a total cost of Rs.18,000 per line. A two-hop DECT link is used to provide connection to the subscriber. One link is between WS and RBS, whereas the other link is between RBS and CBS. Both RBS and CBS use high-gain directional antennas, and are mounted on towers, making a 25 km link possible. The 5 km maximum link distance due to the guard-time limitation of DECT is overcome by use of auto-ranging and timing adjustment [11]. This technique is used in the RBS to support a 25 km link, and to enhance the CBS range to 10 km.

Finally, efficient transmission of packet-switched data on a circuit-switched network is being ensured by the corDECT system, by codeploying a RAS with DIU. Data calls over corDECT are handled differently at DIU from voice calls. The DIU directs an Internet data call from a wallset to the RAS on one 64 kbps slot of the E1 interconnection. The RAS concentrates Internet data from different subscribers and sends them on one or more shared 64 kbps channels set up between RAS and ISP, via DIU and PSTN.

Thus approach differs from the more generic one discussed in Section 2.4 in that an Internet calls from WS does not enter the PSTN at all. Only the multiplexed data on the few shared 64 kbps channels traverses through the PSTN. The data "calls" from WS to RAS terminate in the Access Network itself.

The TeNeT group of IITM along with Vembu Systems (Pvt.) Ltd., Chennai, and Midas Communication Technologies (Pvt.) Ltd., Chennai, has also taken up the development of a cost-effective Fibre Access Network. Designed in accordance with the scheme discussed in Section 2.3.3, the Fibre Access Network again uses a new approach with an aim to provide, apart from the conventional POTS service, large-scale Internet connectivity at a cost affordable in India. As shown in Fig.6, the N-ISDN and HDSL physical layers are exploited in the short copper loop between the RT and subscribers. These relatively high-speed digital links carry both voice and data. The digitised voice signals are directed by the Access Server (AS) towards the RT and then to the Main Exchange. However, the Internet data is separated and passed on to a built-in RAS. After concentrating the Internet data from multiple subscribers, the RAS feeds it to the ISP, via the FAN and PSTN, either on leased lines or on dial-up circuit-switched lines. The subscriber terminal provides multiple telephone sockets and an ethernet interface. The result is one of the most cost-effective means of providing medium and high-speed permanent Internet connections on a wide scale.

Today, the cost of providing Plain Old Telephone Service (POTS) using this FAN is around Rs.9,000 per line. Further, the high-speed permanent Internet connection costs an additional Rs.8,000.

The TeNeT Group of IITM along with Banyan Networks (Pvt) Ltd., Chennai, is in the process of developing a whole range of Remote Access Switches and Access Servers, including those tailor-made for the corDECT Wireless Access System, and the Fibre Access Network as described above. It is also developing a RAS with built-in digital modems to provide Internet connectivity to existing POTS subscribers. In all the products, emphasis is on low cost while maintaining high functionality. The additional cost of the RAS amounts to no more than Rs.600 per Internet subscriber.

Network Management System (NMS) software is today being developed in India by a large number of telecom and computer networking software companies. The capability exists for developing a complex NMS for a large integrated network. Similarly, a number of Indian companies are now developing customer-care and billing systems for clients world-wide. It is possible to leverage their knowledge to customise the NMS and billing software to suit the specific requirements of a new telecom network based on the emerging technologies described in this article.

If a hundred million or more telephones and twenty five million or more Internet connections are to be installed in India, and the service is to be economically viable, the investment per line has to be less than half of what it has been till recently. Recent technological advances in wireless and fibre-based access technologies, open interface standards, high-speed digital transmission on the copper loop, and Internet remote access switches have made this cost reduction feasible. It is possible to install a complete network in India today at per-line cost of Rs.18,000. This would require use of Technologies developed in India over the past few years and will require a continuous striving to come up with new products to add features and further reduce costs. Yet this network would enable one to have a moderate to high-bit-rate Internet access at low tariff. Also, the network can provide high quality service even in small towns and rural areas, something which was considered too expensive so far. Electronics can do for telecommunications in India what it has already done to make personal computers less expensive and more broad-based. What is required is a will to go ahead, belief in Indian capabilities and removal of all bureaucratic hurdles that may come in the way.