T ECHNOLOGIES FOR 4G MOBILE C OMMUNICATIONS AND FOR 3G AND BEYOND HARRI HONKASALO, KARI PEHKONEN, MARKKU T. NIEMI, AND ANNE T. LEINO NOKIA Wide area coverage al GSM (MAP) HSCS 15.2 kb/s 30 khz TDMA (IS-41) CDPD 43.2 kb/s PDC/PDC-P 14.4 kb/s cdmaone (IS-41) 76.8 kb/s The air interface was initially designed to support a wide variety of services with different QoS requirements having maximum bit rate of 2 Mb/s. In order to satisfy the future service and application needs several technical enhancements are being studied and standardized for in 3GPP. 1 ABSTRACT The air interface was initially designed to support a wide variety of services with different QoS requirements having a maximum bit rate of 2 Mb/s. In order to satisfy the future service and application needs several technical enhancements are being studied and standardized for in 3GPP. Even with evolved, there is a need for another public wireless access solution to cover the demand for data-intensive applications and enable smooth online access to corporate data services in hot spots. This need could be fulfilled by together with a high-data-rate cellular system. offers an interesting possibility for cellular operators to offer additional capacity and higher bandwidths for end users without sacrificing the capacity of cellular users. The evolved air interface will provide better performance and higher bit rates than basic, based on first releases of the specifications. Eventually, evolution may not be the answer to all the needs, and some revolutionary concepts need to be considered. However, before some future wireless system can be regarded as belonging to 4G it must possess capabilities that by far exceed those of 3G systems like. Judging from an application and services point of view, one distinguishing factor between 3G and 4G will still be the data rate. We could define that 4G should support at least 100 Mb/s peak data rates in full-mobility wide area coverage and 1 Gb/s in low-mobility local area coverage. Other possible characteristics of 4G need to be further studied. INTRODUCTION Wideband code-division multiple access () was initially proposed and engineered with a vision that already has shown its futureproofness. was designed to be a highperformance system able to support future applications requiring simultaneous transmission of several bitstreams that require individual quality of service (QoS). The original design choice seems to be well aligned with the future, where all applications and services can be carried over IP networks using IP protocols. This trend favors new applications where mobile users have several parallel ongoing sessions based on one or several applications. At the same time is already developing beyond original 3G technology targets, far outperforming any other wireless technology. Furthermore, the possibility to complement coverage and capacity wireless LAN () solutions will be discussed briefly. Even though is exceeding its initial capability targets, there is still a need for a quantum leap in air interface development in the longer term. The quantum leap can be seen as the fourth generation (4G). What this quantum leap is and when it could happen will be briefly discussed. EVOLUTION BACKGROUND: DEVELOPMENT FROM 2G TO 3G Before going into technical solutions in 3G evolution, it is most essential to understand the need for such evolution and essential differences from what the original 3G system can offer. It is important to realize that most often when different generations are discussed (e.g., 2G and 3G), people are referring to major changes in air interface standards. This actually is a good approach, because both core network and applications are developing at their own speeds, and we cannot really see clear differences between generations anymore. Looking back to 2G Global System for Mobile Communications (GSM) evolution and first 3G systems, some essential differences can be seen in the air interface implementation. As an example, one key feature has been support of multiple simultaneous services with different QoS parameters. Another key development has been in data rates, where original GSM could not efficiently support many user needs (e.g., email downloading). However, GSM capabilities have evolved to being close to original targets when is launched. These kinds of new system characteristics, such as support of simultaneous services or increased data rate, first need to be understood before the technical solutions are decided. DEVELOPMENT OBJECTIVES: FOR 3G EVOLUTION The development of 3G will follow a few key trends, and the evolution following these trends will continue as long as the physical limitations or backward compatibility requirements do not 14 1070-9916/02/$17.00 2002 IEEE
force the development to move from evolution to revolution. The key trends include: Voice services will also stay important in foreseeable future, which means that capacity optimization for voice services will continue. Together with increasing use of IP-based applications, the importance of data as well as simultaneous voice and data will increase. Increased need for data means that efficiency of data services needs to be improved as well as delay, and average and peak user data rates. When more and more attractive multimedia terminals emerge in the markets, the usage of such terminals will spread from office, homes, and airports to roads, and finally everywhere. This means that high-quality high-data-rate applications will be needed everywhere. When the volume of data increases, the cost per transmitted bit needs to decrease in order to make new services and applications affordable for everybody. The data rate trends are summarized in Fig. 1. The other current trend in Fig. 1 indicates that in the 3G evolution path very high data rates are achieved in hot spots with rather than cellular-based standards. EVOLUTION ANSWERS TO EXPECTED TRENDS The evolution view from release 99 and release 4 (March 2001) to beyond 3G can be seen in three phases. The three phases are described in detail below. In Third Generation Project Partnership (3GPP) standardization phase 1 is already in progress; the other two phases try to highlight future potentials, but such development has yet to be seen in 3GPP standardization. Phase 1: High-Speed Downlink Packet Access In the first phase, the peak data rate and throughput of downlink for best effort data will be greatly enhanced when compared to release 99. In March 2000, a feasibility study on high-speed downlink packet access (HSDPA) was approved by 3GPP [1]. The study report was released as part of release 4, and the specification phase of HSDPA was completed in release 5 at the end of 2001 [2]. The feasibility study focused on defining a high-speed downlink shared channel (HS-DSCH) that inherits many of the features of the DSCH defined in release 99. The main proposed technical enhancements of HS-DSCH include [2]: Adaptive modulation and coding (AMC) Fast hybrid automatic repeat request (FHARQ) Fast cell selection (FCS) AMC is a radio link adaptation technique where the modulation order and channel coding method are varied according to the quality of the received signal [2, 3]. AMC is somewhat sensitive to measurement errors and delays; therefore, FHARQ has been proposed to provide implicit link adaptation to instantaneous channel conditions [2 4]. FCS has also been proposed to potentially decrease interference and increase the capacity of the system. Using FCS, the mobile terminal indicates the best cell to serve it on the downlink through uplink signaling. Thus, while multiple Mobility Vehicular Pedestrian Stationary 0.1 release 4 cdma2000 1X, EDGE = evolved 2G Figure 1. Data rate trends. HSDPA release 5 cells may be members of the active set, only one transmits at any time [2, 3]. As a result the peak data rate of HS-DSCH will be about 10 Mb/s; based on preliminary simulation results the throughput of a cell/sector will be roughly doubled when compared to release 99. It is anticipated that not all of the proposed technical enhancements will be standardized in phase 1 (i.e., release 5). Release 5 will include some very basic solutions like AMC and FHARQ. Also, from the time-division duplex (TDD) development viewpoint, high data rates can be seen as a necessity. Especially when we think about usage of TDD in the office environment, being competitive with frequency-division duplex (FDD) mode and other indoor solutions is crucial. Thus, HSDPA for TDD mode in release 5 will include some basic technical improvements. Among the interesting research topics are multiple-input multiple-output (MIMO) diversity techniques, which are also studied as part of HSDPA [2, 5]. They can potentially improve system performance quite considerably, as shown in Fig. 2, which depicts the channel capacity based on [6] for different Tx/Rx antenna configurations. Due to implementation complexity they may become reality a little bit later than the other techniques proposed for HS-DSCH. Therefore, from a timing perspective, MIMO techniques could be more relevant for phase 2 or 3 of evolution. Although HSDPA is raised here, 3GPP release 5 work includes and will include many other work items that contribute to 3G development. Phase 2: Uplink High-Speed Data, High-Speed Access for TDD Although the main emphasis in air interface optimization can be seen in the area of downlink high-data-rate support, the uplink also needs attention. Enhanced data rates in the uplink will benefit the end user (e.g., in file 1 10 100 Data rate (Mb/s) 15
30 25 20 15 10 5 0 0 5 Figure 2. Channel capacity vs. number of antennas. 10 8 2 4 2 1 2 1 1 2 8 2 1 transmission or when such office applications as NetMeeting are used). Also, an optimized uplink can be designed to support lower terminal output powers. Additionally, in phase 2 further improvements of HSDPA for both FDD and TDD modes will be seen. This could include the above mentioned FCS and MIMO techniques if proven, performance and implementation points of view, feasible and useful. Phase 3: Capacity Improvements in Uplink and Downlink, and Further Data Rate Enhancement Some of the foreseen air interface technologies become mature in a timeframe that may be unacceptable for the proposed phases 1 and 2 of development. It is obvious, however, that further enhancement will be introduced in later phases. As the demand for very high data rates grows, we can expect the need to further enhance data rates up to significantly above 10 Mb/s. Increase of spreading bandwidth in new frequency allocations could be one answer to the technology challenge. Standardization Timeframe Phase 1 of evolution was completed in 2001. For later phases no approved workplan exist. The approximate schedule for phase 2 could align with following phases of 3GPP releases (e.g., mid-2003 could be the right timeframe). Phase 3 could again take place a couple of years after phase 2, depending on market demand and spectrum availability. OTHER TRENDS 2 4 2 2 15 20 25 The other major trend is the development of picocell and personal area network technologies (e.g., and Bluetooth) for office, public, and home indoor solutions. Current products are able to provide bandwidths up to 11 Mb/s. The next-generation wireless LAN products, based on recently approved standards (IEEE 802.11a and HIPERLAN/2), will offer bandwidths up to 54 Mb/s [7, 8]. In the past, s have been mostly used as a wireless replacement for wired LANs in the office environment. Recently, mobile business professionals have increasingly been looking for an efficient way to access corporate information systems and databases remotely through the Internet backbone. However, the high bandwidth demands of typical office applications (e.g., large email attachment downloading) often calls for very fast transmission capacity. Furthermore, certain hot spots like airports and railway stations are natural places to use the services. However, in these places the time available for information download is typically fairly limited. In light of the above there is a clear need for a public wireless access solution that could cover the demand for data-intensive applications and enable smooth online access to corporate data services in hot spots. Together with high-data-rate cellular access, has the potential to fulfill end user demands in hot spot environments. offers an interesting possibility for cellular operators to offer additional capacity and higher bandwidths for end users without sacrificing the capacity of cellular users, since s operate on unlicensed frequency bands. Furthermore, solutions exist that enable operators to utilize the existing cellular infrastructure investments and well established roaming agreements for network subscriber management and billing. The TDD component of the Universal Mobile Telecommunications System (UMTS) Terrestrial Radio Access Network (UTRAN) is also optimized for hot spot usage on unpaired TDD bands. On one hand, the achievable end user data rates are lower than in systems; on the other hand, the achievable cell sizes are larger. In addition, the cost savings in dual mode between FDD and TDD are significant, and may support implementation of TDD in areas where dual mode with is seen as essential. DEVELOPMENT IN ITU Also, the International Telecommunication Union (ITU) is working on systems beyond 3G. So far the work has concentrated on looking at the objectives for beyond 3G systems. One important aspect is that new spectrum is also needed before beyond 3G systems can be fully deployed. This development is covered in [9]. BEYOND 3G: 4G As mentioned earlier, services, applications, and even the core network are evolving at high speeds, and distinguishing different generations is not really possible anymore. The evolution, and sometimes revolution, is a very significant trend, but in this article 4G is seen as a revolution of the air interface rather than a new phase of evolution. The other major trend is that access methods will be less tightly coupled to the network. This also confuses generational thinking (Fig. 4). After a certain point, evolution is not no longer an answer to air interface development, and revo- 16
200 khz Evolution of 3G Wide area coverage GSM (MAP) HSCS 15.2 kb/s 30 khz TDMA (IS-41) CDPD 43.2 kb/s PDC/PDC-P 14.4 kb/s cdmaone (IS-41) 76.8 kb/s GPRS 170 kb/s 1.25 MHz EDGE 473 kb/s cdma2000-1x 307. kb/s 5 MHz EDGE Ph.2 GERAN 473 kb/s Real-time IP TDD FDD HSPA 2 Mb/s 10 Mb/s T-SCDMA 1XEV - DO, phase 1XEV - DV, phase 2.4 Mb/s 5.4 Mb/s 2 Mb/s High-speed downlink packet access Future wireless Local 802.11b 11 Mb/s HiperLAN2 54 Mb/s IEEE 802.11a 54 Mb/s Harmonized HL2- IEEE 802.11a standard 2000 2001 2002 2003 2010-> Figure 3. The path toward 4G from a radio perspective. The time axis shows the estimated launch times of the actual systems. lutionary concepts must also be considered. Figure 3 illustrates the evolution of 2G/3G cellular and standards and the revolutionary step toward future wireless systems. GSM evolution will continue in parallel with. In the United States, cdma2000-1x will be followed by 1XeV-DO (high-bit-rate data only) and 1XeV- DV (high-bit-rate data and voice) standards. Looking at development in the Internet and applications, it is clear that the complexity of the transferred content is rapidly increasing and will increase further in the future. Generally it can be said that the more bandwidth is available, the more bandwidth applications will consume. In order to justify the need for a new air interface, targets need to be set high enough to ensure that the system will be able to serve us long into the future. A reasonable approach would be to aim at 100 Mb/s full-mobility wide area coverage and 1 Gb/s low-mobility local area coverage with a next-generation cellular system in about 2010 in standards fora. Also, the future application and service requirements will bring new requirements to the air interface and new emphasis on air interface design. One such issue, which already strongly impacts 3G evolution, is the need to support IP and IP-based multimedia. If both technology and spectrum to meet such requirements cannot be found, the whole discussion of 4G may become obsolete. SPECTRUM ISSUES WRC2000 already identified new spectrum for IMT-2000 systems. The ITU identifications at 2 GHz frequency range can be seen in Fig. 5. In addition to 2 GHz identifications, WRC2000 also identified parts of the 806 960 MHz band that already have primary mobile allocations to IMT-2000. Intranet Internet 2G/3G LAN Bluetooth xdsl Figure 4. Coupling of air interface technologies to the network. The demand for even higher data rates and potential need for wider bandwidths in cellular evolution raise even further questions on spectrum needs. However, before this need can be defined, a much better idea/vision of the next generation will be needed. For example, if is combined with, the additional spectrum is already the current spectrum, and something more is needed to justify even higher spectrum requirements. There are many discussions ongoing, for example, about reallocating analog TV bands for mobile systems. On the other hand, administrations seem to prefer more flexible air interfaces where globally harmonized spectrum would not be needed. ISDN POTS TDMA GSM GPRS CDMA 17
ITU identifications S5AAA S5388 S5388 S5AAA 1700 1750 1800 1850 1900 1950 2000 2050 2100 2150 2200 2500 2550 2600 2650 MHz Figure 5. ITU spectrum identification: S5.388 are the WARC 92 identifications, and S5.AAA are the additional WRC 2000 identifications. CONCLUSIONS The air interface is seen to develop far beyond its initial capabilities to satisfy future service and application needs. systems are seen to complement -based cellular evolution in hot spots in development beyond 3G. The other technology to be considered for hot spot coverage is UTRAN TDD mode, which is well harmonized with FDD and in such a way facilitates cost-efficient dual mode terminal design. When looking at development of services, applications, or core networks, development especially in applications is much faster than traditional generation thinking assumes. This development will happen in an evolutionary way without clear generations. That is why here we consider quantum leaps in air interface development as different generations. The next such quantum leap will lead us to 4G. This thinking is well in line with development from 1G to 2G and 3G. People clearly refer to air interface standards when referring to these generations. 4G needs to be something that 3G evolution cannot do. Looking at the complexity of application contents and development of such contents, going toward even higher data rates and availability of high data rates everywhere is a trend. Thus, one distinguishing factor between 3G and 4G will be the data rates. We assume that 4G should support at least 100 Mb/s peak rates in full-mobility wide area coverage and 1 Gb/s in low-mobility local area coverage. There will be other characteristics for 4G, but at this point in time the requirements for 4G need further studies (including market studies). ACKNOWLEDGMENTS This article is based on previously published material from 3Gwireless 2001 organized by Delson Group (http://www.delson.org). REFERENCES [1] Motorola, Work Item Description sheet for High Speed Downlink Packet Access, TSGR#7(00)0032, March 13 15, 2000, Madrid, Spain, p. 3. [2] 3GPP TSG-RAN WG2, UTRAN High Speed Downlink Packet Access (release 4), TSG-R2 TR 25.950. [3] Nokia, Considerations on High-Speed Downlink Packet Access (HSDPA), TSGR1#14(00)0868, July 4 7, 2000, Oulu, Finland, p. 9. [4] Nokia, Text proposal on HARQ for HSDPA TR, TSGR1#17(00)1369, 21 24, Nov., 2000, Stockholm, Sweden, p. 4. [5] G. J. Foschini and M. J. Gans, On Limits of Wireless Communication in a Fading Environment when Using Multiple Antennas, Wireless Pers. Commun., vol. 6, no. 3, Mar. 1998, pp. 311 35. [6] E. Telatar, Capacity of Multi-Antenna Gaussian Channels, Technical Memorandum, Bell Labs, Lucent, Oct. 1995, published in Euro. Trans. Telecommun., vol. 10, no. 6, Nov/Dec 1999, pp. 585 95. [7] ETSI, Broadband Radio Access Networks (BRAN); High Performance Radio Local Area Network (HIPERLAN) Type 2; Requirements and Architectures for Wireless Broadband Access, TR 101 031 (1999-01), v. 1.1.1. ETSI, 1999. [8] ETSI, Broadband Radio Access Networks (BRAN); High Performance Radio Local Area Network (HIPERLAN) Type 2. ETSI Technical Specification, Physical Layer, v.1.2.2, 2001. [9] D. McFarlane, Enhancing the Capabilities of 3G Systems, IEEE Int l. Conf. 3G Wireless and Beyond, 30 May 2 June, 2001, San Francisco, CA. BIOGRAPHIES HARRI HONKASALO received his M.S. from Helsinki University of Technology in 1986. In 1985 he joined Nokia. From 1985 to 1993 he held various research and product development positions in Nokia Mobile Phones in Finland and the United Kingdom, working on GSM and PDC terminals and standardization. In 1994 1995 he was with Nokia Research Center in Finland, responsible for research activities on GSM standards, and in 1996 1997 he was with Nokia Research Center in the United States, responsible for research activities on IS- 95 and cdma2000 standards. From 1998 to 2000 he was head of system research in Nokia Mobile Phones with global responsibility for system research activities. Since 2001 he has been director of IPR for Nokia Corporation with global responsibility for standards-related IPR activities. He has published about 10 papers in international conferences and journals, and holds 13 patent families covering different areas of radio interfaces. KARI PEHKONEN [M] (kari.pehkonen@nokia.com) received his M.S., licentiate in technology, and doctor of technology degrees from the University of Oulu in 1987, 1989, and 1993, respectively. He joined the Computer Technology Laboratory of the Technical Research Centre of Finland in 1987 and was involved in research on parallel programming and parallel computers. During 1989 1990 he was a visiting researcher at the Computer Vision Laboratory of the Center for Automation Research of the University of Maryland, doing research on computer vision algorithms. Upon returning to Finland he continued with the Technical Research Centre, studying further the algorithms developed during his visit to the United States. Since 1993 he has been with Nokia Mobile Phones, first as a research engineer and then holding various managerial positions within the company, doing research on systems and standardization. From 1998to 2001 he was with Nokia Japan, responsible for ARIB standardization activities. Since the beginning of 2001 he has been head of system research at Nokia Mobile Phones with global responsibility for system research activities. He has published about 20 papers in international conferences and journals and holds 9 patents covering different areas of radio interfaces. His current research interests include the system aspects of radio access networks with a special interest in L1 solutions. ANNE-TUULIA LEINO received her M.S. degree from the University of Technology, Espoo, in 1991. She joined the Telecommunications Administration Centre, Finland, in 1991, working on spectrum topics related to the regulation of public mobile networks. She changed to Nokia Networks in 1997 to work as a spectrum expert covering frequency arrangements related to IMT-2000 networks. MARKKU NIEMI received his M.S. degree from Tampere University of Technology in 1995. He joined Nokia Mobile Phones in 1995 and currently is senior manager at Nokia Mobile Phones responsible for standardization, regulatory matters, and research cooperation. Prior to his current position he held various management positions inside Nokia Mobile Phones. He has published about 10 papers in international conferences and journals, and holds several patent applications. 18