By Stanley Chia

For many industry observers, it’s clear that CDMA will offer the best platform for third generation systems. But there is more than one CDMA path available. For many established operators already using cdmaOne technology, the ideal third generation approach would involve an evolution from the existing standard. Can cdmaOne deliver?

CDMA has long been recognized by cellular radio development engineers as a viable approach for delivering feature-rich mobile services. For instance, back in the mid 1980s when GSM first selected its radio interface, two out of the five candidate system proposals, CD900 and MATS-D, were based on some form of code division spectrum spreading. However, due to factors linked with technology maturity and implementation cost, the CDMA elements of these proposals were not incorporated in the final GSM Specification.

While the Europeans decided to proceed down the TDMA path, the momentum of CDMA continued to build rapidly in North America in the early 1990s with Qualcomm leading the way to create a new cellular system based on the IS-95 standard. As cdmaOne, this standard is rapidly gaining popularity worldwide. Recognizing the importance of CDMA, the European Commission and various industries soon responded by launching a project to develop a CDMA based test-bed for third generation mobile systems under the European-wide RACE II (Research into Advanced Communi-cations for Europe - Phase 2) Program. This has established Europe’s interest in adopting some form of CDMA for evolving to a third generation mobile system. Likewise, in Japan, the conclusion was reached by NTT DoCoMo that its third generation mobile system should also be CDMA-based. Thus, CDMA has firmly emerged as a global focus of the third generation mobile system standardization efforts.

BENEFITS TO CUSTOMERS
In the transition from first generation analog systems to second generation digital systems, customers were able to enjoy better voice quality with the evolution of vocoding technologies as well as greater privacy, longer battery life and a much larger portfolio of data and supplementary services. Likewise, in the evolution to third generation systems, the criteria currently driving developments are:
• Scaleable, flexible multimedia services including Internet access
• High service quality
• Lower cost and improved performance
• Seamless global roaming and mobility
With these capabilities, third generation mobile systems hold the promise of increasing service choice and flexibility to the customers. In particular, customers will be able to tailor their own personal profile for accessing innovative and interactive multimedia services as well as being able to access individual media components. Service on demand will also be possible, with high quality entertainment services and on-line surfing delivered to terminals. Public and business information services, Internet access, virtual education facilities, video telephony, road transport telematics as well as telemedecine and improved emergency services are further examples of the potential benefits. As a whole, third generation mobile systems will enable communications and information provision at the right place and at the right time without the constraints of bandwidth and distance of second generation systems.

Third generation mobile systems are expected to provide high data throughput and coverage as shown schematically in Fig 1. In the nearer term, it is envisaged that a data rate of no more than 2 Mbit/s will be required to support the above services and, in practice, an average of 144 kbit/s should satisfy most of the perceived needs. Support of 2 Mbit/s data rates refers more to the peak data rate. Clearly, the data rate which can be supported in different environments at mobile speeds will be different. Specifically, for vehicular speeds, the peak data rate will be up to 144 kbit/s. For pedestrian speeds, the corresponding data rate will be 384 kbit/s. For an indoor office environment, however, the transmission can be up to 2 Mbit/s.
By contrast, in countries where basic telephone infrastructures are less well developed, third generation mobile systems could be used to provide better basic services at a lower cost and a better performance.
With the advent of feature-rich third generation systems and voice quality expected to be at least equal to fixed, mobile operators may be well positioned to introduce competition into the fixed wire market, thus benefiting end users, an important consideration in the development of third generation mobile systems. Service portability and transparency when roaming from one country or network to another will also be of great importance.

In summary, third generation mobile systems have two broad service requirements: multimedia and high-speed data for a high end service offering and low-cost voice for a general consumer application at the low end. Fixed wireless can, indeed, be viewed as a service category for the developing world. However, high service quality and low cost are common to both developed and developing countries.

MARKET SIZE
With a strong projected demand for telecommunications driven by the anticipated uptake of wireless multimedia services, the size of the third generation mobile market could be quite large for both suppliers and operators. It has been forecast that the mobile market could rival the fixed market by 2010 (see figure 2). The European UMTS Forum forecasts that, by end 2010, the global number of mobile users could reach 1.7 billion. This figure, however, merely represents a penetration of 20% of the world’s population.

There is some uncertainty about the penetration rate. For example, the UMTS Forum forecast that more than 45% of mobile users will have high-speed data uptake by 2010 whereas other, low end forecasts have shown that this could be less than 10%. On the other hand, the voice market is also expected to grow and could be perceived as an opportunity for third generation technology provided it can offer sufficient cost and performance advantages.

The highest growth region will be in the Asia Pacific rim with some 40% of the world’s future subscribers. The second highest is North America and Latin America, which collectively constitute an additional 33%. Europe is expected to have 25% of the subscribers, with Africa and Middle East at 1% to 2%.

STANDARDIZATION
Worldwide standardization of third generation mobile systems has been in progress within the International Telecommuni-cation Union (ITU) for some time under the IMT2000 initiative. In addition, regional activities in Japan and Europe have gained considerable momentum over the past few years. Specifically, the Japanese have been active in striving for a launch by year 2000, while the Europeans are currently working towards selecting a radio access method by the end of this year. In North America, the CDMA Development Group is making a concerted effort to evolve cdmaOne to a third generation system. The overall timeline for the ITU is shown in Fig 3 together with the European and Japanese timetables.

Japan. With limited overseas markets for their second generation PDC (Personal Digital Communications) system so far, Japan’s Ministry of Post and Telecommunications (MPT) wanted the new third generation standard to be more widely adopted across the world. The MPT’s timetable for third generation system introduction suggests a call for operators in 1999. In the early part of the year, the first issue of the standard will be made available and license awards will begin during the latter part of the year. Commercial deployment is expected to be round 2001. In recent months, NTT DoCoMo has invited infrastructure and handset vendors to participate in its wideband CDMA development. An experimental system is expected to be operational by early 1998.

As shown in Table 1, NTT DoCoMo proposes to use different modulation and coding from cdmaOne (IS-95) with different Layer 2 and 3 signaling, different handoff algorithm and asynchronous base stations. NTT DoCoMo has not established a requirement for backward compatibility with any existing second generation system or analog standard. In addition to the adoption of INAP (Intelligent Network Application Part, Signaling System Number 7) protocols and ISDN (Integrated Services Digital Network) call models, new protocols will be established for services and mobility. Interoperability with existing second generation systems is proposed to be provided through gateway functions.

Europe In Europe, five contending concept groups for access technology have been proposed as candidates for standardisation. These include:
Wideband CDMA — Frames Mode 2: Fujitsu, NEC, Panasonic
OFDM (Orthogonal frequency division multiple access) - Sony, Telia and lucent Wideband TDMA — Frames Mode 1 without spreading
Joint detection TDMA/CDMA — Frames Mode 1: Swiss Telecom PTT
ODMA (Opportunity driven multiple access): Vodafone and Slabu R&D
Frames is part of the European-wide research programme with key industrial partners including Ericsson, Nokia and Siemens. A comparison of the European wideband CDMA approach is shown in Table 1. Overall, the European timescale is that a radio interface will be selected from one of the five concept groups by the end of 1997, regulators will call for operators by mid-1998 and award licenses in 1999, and the standard will be ready around 2000.

EVOLUTION FROM cdmaOne
While the evolution path to a third generation system is dependent on many factors, for operators with cdmaOne properties, an evolution path which allows coexistence and integration of both narrowband and wideband CDMA would be preferred. However, there is always the challenge to the industry of preserving backward compatibility and meeting the customer ideal of global roaming. As shown in Table 2, proposals are already emerging from leading infrastructure vendors including Qualcomm, Lucent, Motorola, Nortel, Hughes Network Systems, Nokia and Samsung to make the transition from cdmaOne to a third generation system feasible.
Enhancing cdmaOne. Utili-zing the statistical multiplexing capability of CDMA, the enhanced cdmaOne concept is to retain the cdmaOne channels as fundamental channels and add supplemental channels which can be operated in conventional CDMA mode or in bursty broadband mode. In the latter broadband packet mode, the channel is directed to specific users based on the packet address. Thus, the supplemental channel becomes a packet access scheme; a number of supplemental channels can be implemented simultaneously depending on the traffic offered by the users.
The main advantage of this approach is that a frequency reuse of one can be preserved. Another advantage is that it can maintain the voice/data rate and carry Internet protocol without queueing. A mobile can be directed to a specific supplemental channel without the need to consume additional battery power. While the fundamental channels are always in soft handoff mode, the supplemental channels need not be so. This simultaneous channel operation will achieve some fundamental capacity gain at the expense of requiring to optimise the link regularly. It is estimated that by this approach, the supplemental channel can support a peak data rate of 307 kbit/s.
It is also possible to enhance the number of fundamental channels. One way to do this is to add more orthogonal Walsh channels; an alternative is to make use of the QPSK modulation to put different information in the in-phase and quadrature (I&Q) domain. While the former is better for backward compatibility, the latter can prevent phase and amplitude from being disturbed due to unequal power levels. More work is planned in the CDG to further evaluate and recommend the optimum approach.
A wideband carrier approach. Among the proposals, one approach is to form a 3.6864 Mbit/s wideband channel to coexist or overlay on existing 1.2288 Mbit/s cdmaOne channels to allow backward compatibility and transition flexibility. This is shown schematically in Figure 4. The reason for requiring 5 MHz to house the 3.6864 MHz channel is because of the guard band requirement. For example, with a 15 MHz allocation, either eleven cdmaOne carriers or two cdmaOne carriers with three wideband carriers can be implemented.

Proposed by Motorola under the banner of "Wideband 95", the main objective of this scheme is to improve the forward link capacity and the reverse link range. The technology can be deployed in any band and is, therefore, fully backward compatible with existing deployments. It is designed for spreading over a bandwidth of 1.25 MHz, 5 MHz, and10 MHz and may also be 20 MHz to support 144 kbit/s, 384 kbit/s, 512 kbit/s and 2.0 Mbit/s peak data rates, respectively. In theory, the wideband spreading technique can continue to be used to increase the spreading bandwidth with increasing receiver complexity. The limit will be the point at which there are too many multipaths of which some are too low in amplitude to be fully tracked by a rake receiver.

A number of potential technical challenges could result from this approach to which solutions have to be provided. For the reverse link, the linearity of the power amplifier may be an issue. By contrast, for the forward link, the number of rake fingers has yet to be optimized. It has been proposed that a pilot-assisted coherent reverse link be used to improve range and a fast forward link power control algorithm be incorporated to improve capacity.
The question of directly overlaying a 3.69 Mbit/s Wideband 95 channel over 1.23 Mbit/s cdmaOne channel is a complex one. At the time of writing, industry experts believe that, for the forward link, overlay operation could lead to channels losing orthogonality and the possibility that power may have to be increased. As for the reverse link, since the channel is not orthogonal, this is not a problem. Direct overlaying will have a capacity degradation effect, especially for the narrowband system of up to 30%. Since the wideband channel is not completely orthogonal to the narrowband channel, it will appear as noise to the latter. However, the industry is exploring ways to mitigate these potential effects and to improve the performance in simultaneous channel operation.

An alternative which is not explored in the standards today is to apply some form of time division multiplexing to partition the spectrum utilisation of the wideband and narrowband channels to ensure orthogonality (see Figure 5). Instead of using a fixed assignment for all channels, the spectrum can be used on a time-shared or statistical basis. Depending on the demand and priority, the wideband transmission is suppressed during the narrowband transmission and vice-versa. This could permit simpler coexistence and ease the migration path.

A multicarrier approach. By contrast, the multicarrier approach is to directly concatenate three 1.2288 Mbit/s channels together to provide a total bit rate of 3.6864 Mbit/s for the forward link. This is shown in Figure 4. With a three-carrier configuration, a data rate of up to 153 kbit/s using QPSK modulation can be achieved. This app-roach is designed to reuse the multiple carrier architecture of the existing base station infrastructure. The efficiency is expected to be 5% to 10% below a wideband spreading scheme. For the reverse link, a wideband carrier spanning a 5 MHz spectrum is proposed. This is designed to ensure that handsets are not required to transmit multiple carriers and to take into account linearity issues.
The proposal is also supposed to be scaleable between 5 MHz and 15 MHz. Proponents of this approach claim that it makes it easier to achieve backward compatibility with the existing deployment. In addition, the use of multiple replicated waveform also helps to simplify the signal processing, as complex wideband waveform generation could be relatively difficult to achieve.

It should be noted that handsets for the multicarrier approach would be more complex than those used in the wideband approach as the former have to demodulate three carriers simultaneously. In addition, synchronised power control could also be an issue.

While the worldwide third generation mobile system standardization effort is addressing specific greenfield frequency bands, it is important to realize that there is no fundamental restriction to the evolution from any frequency bands. For example, the migration from the existing cdmaOne to W-95 can be achieved with a similar procedure to that involved in migration from AMPS to cdmaOne. Of the two key approaches discussed above, the direct sequence approach will be better for greenfield implementation, while the multicarrier approach is expected to be better for migration with backward compatibility. The exact magnitude of this difference will be studied extensively before any final decisions are made. However, if an operator has a 5 MHz allocation and if at least 1 x 1.25 MHz is already in use and is busy, the implementation of either approach could still be challenging.

The trade-off between the two is in system efficiency and handset complexity. However, at this point, not enough performance differences between the two approaches can be noted. For the direct sequence approach, it is better to have the 1.25 MHz channel residing outside the wideband channel as the transmitter radio front end for the two channels will be different and will perform better if the two channels are on separate carriers. If an operator is spectrum constrained, derating the voice channel capacity and overlaying the 3.6 MHz data channel on top could be considered. In this case, more base sites for the cdmaOne users may be required. If the band has to be cleared prior to wideband services implementation, the technology has to be very cheap to encourage their introduction.

Based on the economics of the two approaches, operators with a larger amount of spectrum allocation would benefit from the wideband approach, while cellular band operators would benefit from the multi-carrier solution. A summary of the comparison is shown in Table 3. The above discussion would support arguments on where new spectrum is to be made available for third generation mobile systems. An allocation to existing operators would be beneficial to foster the delivery of these new services to customers.
With both proposals, much can be reused in the transition from cdmaOne to a third generation mobile system. For instance, the reuse of base site equipment is possible. For the 1.25 MHz channels, the existing cdmaOne cell sites are directly applicable and thus provide backward compatibility. In addition, the A-interface, signalling and the channel card, frame format, IS41 and handover procedures can all be reused in the evolved systems.

ENABLING
TECHNOLOGIES
In addition to the multiple access techniques, many other enabling technologies are required to provide the full capabilities of a third generation system. In particular, progress in radio, network and terminal technologies are of vital importance to ensure the success of the evolution.
Networks. In the near term, operators are likely to stay with 64 kbit/s circuit switching as well as packet switches but continue to evolve the platforms to adopt new information technology architectures and support delay-sensitive services simultaneously. TCP/IP will continue to be the network layer protocol with ATM physical and link layer; this may extend to the desktop for both mobile and fixed applications if ATM is successful in penetrating corporate networks all the way to the office desks. Over time, operators can gradually migrate all switching capability to ATM-based switch fabric and adopt more sophisticated broadband ISDN signalling protocols.
In the longer term, a possible objective could be to evolve the entire access network switching fabric to be ATM-based connecting base stations of different technologies linking directly to applications (SMS, GPRS, voice mail etc.) and resource (HLR, MSC) nodes. The ATM switch would link into the core network for access to Intelligent Network based services supported by an SS7 signalling network. It is expected that the architecture will be better scaleable than today’s processor and memory-based approach.

This platform will lead to a departure from today’s direct association of the switch cost with performance and provide a more competitive platform for operator differentiation through the use of in-house developed or third-party service products. With this platform, operators can capitalise on common and commercialized technologies to reduce service development costs and enable a faster time to market.
Radio. As a technology enabler for the radio subsystem, a low bit rate encoding scheme is a challenge especially when used with compression. The service integrity is expected to be extremely sensitive to the channel quality and thus error control becomes vital.

Other key issues which are receiving close attention from researchers and manufacturers are radio resource management techniques, the round trip delay of codecs, fast wide dynamic range A/D and D/A converters, radio frequency integrated circuits, spectrally efficient linear power amplifiers with up to 75 dB or 80 dB dynamic range, and spatial processing using smart antennas.

In general, the amount of radio frequency processing continues to reduce at the expense of increased digital signal processing. With the maturity of high speed analog-to-digital conversion with 14-16 bit resolution, 70 dB dynamic range and a sampling rate of 30 MHz being available at a reasonably low cost, the development of wideband software radio systems becomes more feasible.

The crossover point between discrete channel approach and block down-conversion approach is getting closer. In the past, the crossover point was at 20-30 TDMA channels; this is now coming down to four to five channels.

While VLSI baseband circuitry can be available at nearly zero cost, leading manufacturers have also pointed out that the radio frequency (RF) component cost has been a challenge in the past and will continue to be so in the future. By contrast, although linear power amplifiers have been expensive and challenging to implement up to now, breakthrough techniques are already emerging which could lead to significantly more cost effective implementation.

Terminals. Third generation system handsets will typically have dual bandwidth, dual frequency with dual resonant antenna and integrated RF front end.

While software downloadable vocoders are feasible, the use of customized as opposed to multi-mode modems is seen by the industry to be the option which makes more economic sense.

As for the handset itself, in the future the handset will simply consist of ASICs, a keyboard, and an antenna connector in an enclosure. Terminal equipment can, however, have many options ranging from being a virtual full client to split client, very thin client, partial client and defeatured client.
With CMOS technology continuing to shrink to 0.18 micron and costing near zero cost per gate, the overall trend in handset development will be an increase in the sampling rate and a loading of more and more software rather than the use of firmware.

Battery technology and power management are expected to be further key elements. The current talk time is for supporting a data rate of 10 kbit/s for a few hours. With a data rate of the order of 100 kbit/s, the same battery will not be able to support long data sessions. Thus, thin client software and non-memory intensive equipment are also key parts of the technology development.

In conclusion then, efforts under way in worldwide third generation system development suggest that CDMA has many advantages that make it attractive in meeting the requirements for high quality broadband service features and characteristics. In particular, we have seen that evolution from cdmaOne is a feasible approach, with technical proposals emerging from leading infrastructure vendors. Differences in frequency bands and details of the vendor proposals suggest that global intersystem roaming may still be as challenging with third generation as it is with second generation mobile systems.

With the parallel effort in Europe and Japan on third generation system standardisation, operators and suppliers should work together toward the ultimate goal of ensuring that customers can benefit from better quality and higher service flexibility of next generation systems whether they are at home or travelling.