[For a copy of this article in PDF format, which displays figures and equations, click here. Requires Adobe Acrobat Reader, free download.]

Multimedia presents a new challenge to mainstream wireless technologies. Will 3G be able to deliver?

Fifteen years ago mobile telephones were an exotic extravagance. Today, as cellular phones, they are often given away as freebies in support of marketing schemes and product promotions. Having become a mainstream voice communications medium, they are poised to take on new challenges, transmitting (fairly) high-speed data, video, and multimedia traffic, as well as voice signals to users on the move. The primary need is fueled by increasingly mobile workforces in every industry. According to the Strategis Group, an average of 15% of the U.S. workforce is classified as mobile (out of the office at least 20% of the time). The technology needed to tackle the challenges is known as third-generation mobile telephony (3G).

Background Access speeds in wireline and wireless environments are defined either as narrowband, (less than 64 kb/s), or broadband, (greater that 64 kb/s). The majority of mobile data offers available to date fall into the narrowband category. This enables simple data push, two-way messaging (SMS) and limited, Internet browsing using the wireless Internet protocol (WAP) or Internet clipping. In Europe, narrowband data access is currently provided over the global system for mobile communications (GSM) networks, using SMS. In Japan, data services operate over NTT's iMode network. In the United States, data services use network overlay architectures such as cellular digital packet data (CDPD) (with a maximum theoretical speed of 19.2 kb/s) or CDMAOne (with a maximum theoretical speed of 14.4 kb/s). There are also a number of dedicated mobile data networks already in existence, namely the Advanced Radio Data Information Service (ARDIS), Mobitex (BSDW) used by Palm.net, and Bell South Mobile Data (RAM) which support paging services, and Metricom's MCDN.

Why 3G? - a consumer view The need for more voice capacity to accommodate more callers reliably and to generate more revenue through increased billable minutes of use (MOU), high-speed data services (384 kb/s mobile and 2 Mb/s fixed) to cater to the needs of an increasingly mobile user community, corporate local network access, wireless Internet access and global roaming, have pushed the limits of existing 2G networks. Proposed 3G implementation will support such needs. The chief requirements of any 3G technology initiative, declared or implied must include:

- Voice quality comparable to that of Public Switched Telephone Network (PSTN)

- A data rate of 144 kb/s for users in motor vehicles moving fast over large areas

- A data rate of 384 kb/s for pedestrians, standing still or moving slowly over small areas

- Phased-in support for 2.048 Mb/s operation office use

- Support of both packet-switched and circuit-switched data services

- An adaptive radio interface suited to the highly asymmetric nature of most Internet communications - a much greater bandwidth for the downlink than the uplink

- More efficient usage of the available spectrum

- Support of wide variety of mobile equipment (phones, handhelds, personal digital assistants (PDAs) etc.)

- Flexible introduction of new services and technologies.

Another significant driver for 3G is the demand for mobile access capacity due to proliferation of mobile Internet. U.S. Internet penetration is growing rapidly and is expected to reach over 65% in the next few years (ADVENTIS), while conservative estimates of U.S. mobile phone penetration predict an increase from 40% today to 50% by 2004 (Strategis Group). In Western Europe, the picture is reversed with Internet penetration of around 45% expected by 2003, and mobile phone penetration of over 65%. However, in both regions, research has shown that mobile phone owners are likely to be Internet users, with 75% accessing Internet on a regular basis (Yankee Group), indicating potential receptivity to mobile data. Mobile data marries demand for connectivity with demand for mobility, and is leading a convergence between Internet and wireless value propositions. Mobile data is expected to reach 15% penetration among U.S. wireless subscribers by 2004 (Strategis Group), with wireless ecommerce growing $21 billion by 2003 (IDC).

Ideally, the third generation should provide seamless personal communications services anytime anywhere.

Why 3G? - a supplier view To an extent, demand for mobile data is also being driven from the supply side. Traditional network operators need to differentiate their offers to offset commoditization of voice traffic and have started to offer value-added data services, although take up of these low data rate services (ranging from 2.4 kb/s to 9.6kb/s) has been limited to date. Content providers, facing increased on-line competition, are also seeking to retain customer share by providing "anytime, any where" access to their sites. To this end both groups are developing mobile portals, or "mortals," providing popular content and Internet access to wireless cellular subscribers.

The requirements of a full-blown 3G system may be somewhat of an overkill for the "average" consumer. Professional users, on the other hand, will demand a broad range of more bandwidth-intensive applications. The Strategis Group estimates that by 2004, 21.5 million business professionals will be subscribing to mobile data services in the United States. Compared to the broader market for basic mobile data services, this represents a niche. However, it is this group that will really drive the demand for high-speed mobile access and pay premium prices based on the value that they derive from these services.

The Path to 3G Ten terrestrial radio transmission technologies (RTTs) that met the minimum performance capabilities of IMT-2000 were submitted to the ITU in June 1998. Code-division-multiple-access (CDMA) is used as the air interface in all of them except for digital enhanced cordless telecommunication (DECT) and universal wireless communication (UWC-136), which are pure time-division multiple-access (TDMA) proposals. Supporters of the different RTTs are going to a great deal of trouble to make them converge into a set of unified standards. Already the technologies are clustering around the three main 2G legacy standards: TIA/EIA-136 and GSM (both TDMA-based systems) and IS 95 (CDMA-based). Since the evolutionary heritage of mobile radio is respected in all three of the convergence efforts, each of the RTTs is likely to be included, to a greater or lesser extent, in whatever converged technologies finally emerge in the standards. Except for the highly regional pacific digital cellular (PDC) system in Japan, all the second-generation investments are protected by one or more radio transmission technologies.

As shown in Figure 1, the paths to the future are many and varied, some adhering to a single 2G technology while others try to be more inclusive of multiple technologies.

Existing mobile data technologies Before moving to a discussion of new 3G technologies, it helps to have an overview of existing 2 or 2.5G technologies that are used today for data services. Limitations of existing technologies are fueling the needs for the new 3G technologies.

- Two-way dedicated data networks Established in the 80s, RAM and ARDIS (now Motient) are examples. Both support research in mo - Blackberry (RIM) interactive pagers and wireless modems for mobile laptop access, operating over private radio networks. The majority of applications are targeted at industries, i.e. transportation, field sales, security etc. They offer e-mail, database access, dispatch-tools and personal information management (PIMs). Both offer wide coverage across North America. However, despite this extensive market presence, theoretical maximum burst data rates offered by ARDIS and RAM are less than 19.2 kb/s and less than 9.6 kb/s respectively. They are inadequate to support high bandwidth mobile applications.

- CDPD CDPD is another first-generation data service providing packet-based transmission over TDMA- and CDMA-based networks, which enable mobile data access at theoretical maximum burst data rates of up to 19.2 kb/s. CDPD supports data services on WAP-based handsets. By introducing packet-based transmission capabilities, more efficient use is made of existing wireless capacity, supporting a large number of subscribers and enabling data services. However, access speeds are low and inadequate to support high bandwidth mobile applications.

- High-speed circuit-switched data HSCSD is an interim GSM upgrade designed to speed circuit-switched mobile data rates. Using HSCSD, data are transmitted across as many as six voice channels downstream, at 9.6 kbp/s, providing a maximum theoretical burst rate of up to 57.2 kb/s, and an average peak rate of 38.4 kb/s, and two upstream channels with a combined theoretical maximum burst rate of 19.2 kb/s. While the use of dedicated circuits ensures a higher data quality than shared packet networks, the capacity is far lower and the ability to support high-speed mobile data services is questionable. As data traffic increases, voice channels are sacrificed, forcing trade-offs in capacity use. The use of multiple channels also implies services need to be priced at eight times voice traffic to generate comparable returns, making data services prohibitively expensive. Only a limited number of GSM operators are planning on HSCSD roll out prior to GPRS roll out.

- CDMAOne CDMAOne is another 2G implementation using spread spectrum technology to support data services over CDMA networks, and forms part of the planned evolution to 3G based on Qualcomm's CDMA 2000 (3xRTT). Sprint PCS is currently offering WAP-based circuit-switched data services using CDMAOne with a theoretical maximum burst speed of 14.4 kb/s, although with upgrades to support packet data, a maximum burst speed of 64 kb/s is predicted in the near future. Under CDMA 2000 phase 1 (1xRTT), theoretical maximum data burst rates are expected to increase to 144 kb/s, although average rates have only reached 28.8 kb/s in the first trials. CDMAOne is being widely deployed by a number of U.S. carriers due to the relative ease and limited cost of upgrading existing networks. However, present data rates do not scale significantly and are currently too low for mobile business applications.

New 3G Technologies - Wideband CDMA One of the most promising approaches to the new, 3G is to combine a wideband CDMA air interface with the fixed network of the GSM. The network supports roaming and is the preferred choice in more than 110 countries, especially in Europe.

The W-CDMA proposal is based on ETSI's UMTS. Called UMTS terrestrial radio access (UTRA), the proposal describes two operating modes: frequency- and time-division duplexing. In the first mode, a physical channel is a unique frequency/code assignment. A user requiring service capable of handling a high data rate could be assigned multiple physical channels. The uplink channel corresponds to a specific carrier frequency, a spreading code, and a relative phase (0 or p/2 radians), which means it can accommodate multiple information streams.

A 10 ms frame structure resides under the assigned spreading code. Each 10 ms frame is divided into 16 slots of 625 ms duration, each of which corresponds to a power control period (there are 1600 power control adjustments per second).

The downlink frame structure in the frequency-division duplexing mode is identical to the uplink structure, in which user and signaling information are time multiplexed within the 625 ms slots. The spreading codes in both uplink and downlink directions have spreading factors that vary from 4 to 256, where the spreading factor is inversely proportional to the required data rate. When the bit-rate to be transmitted in one downlink channel exceeds the maximum allowed, several parallel connections can be established using the same spreading factor.

The TDD mode has a 10 ms frame structure composed of 16 slots, each 625 ms long. Multiple switching points can be set within the 10 ms frames to accommodate asymmetric uplink and downlink circuits. Each 625 ms slot is spread with its own unique code. Two or more 625 ms bursts can be accommodated in each slot, distinguished by its own spreading code. Hard handovers between GSM (2G) and UTRA (3G) systems are supported. The GMSK modulation scheme in GSM is replaced by QPSK.

- CDMA 2000 CDMA 2000 technology was submitted with the interest of protecting the numerous IS95 systems deployed in the United States and Korea. The proposed technology exploits the full ability of the current 2G CDMA systems to accept some 3G features. In fact, the current CDMAOne systems can be viewed as narrowband versions of the fully evolved 3G CDMA 2000 system.

The technology supports handoffs between 2G and 3G systems as well as both FDD and TDD radio techniques. A CDMAOne system can deploy some of the 3G features without increased channel bandwidth, provided that certain signaling details and logical resources within the 1.25 MHz channel are modified to meet the needs of packet radio and asymmetric services. More broadband features may be added later by multiplexing additional CDMAOne channels in increments of 1.25 MHz.

- UWC 136 Seventy percent of the mobile world today uses GSM technology. UWC 136 is a proposal that envisions a purely TDMA standard with no CDMA elements at all - a property that opens up the possibility of deploying advanced features within frequency bands as narrow as 1 MHz. UWC 136 is also important because it represents an evolutionary path for the advanced mobile phone system (AMPS) and 2G IS 136 technologies. There are more AMPS handsets in the world than any other, except for GSM. This 3G technology offers greater data rates than its 2G counterpart, support for asymmetric uplink and downlink channels, and optional packet data support.

Advanced features call for greater channel capacity than the last generation networks can provide. Existing frequency-division multiple access (FDMA) or TDMA systems are confined by regulatory constraints to increasing channel capacity without increasing bandwidth. This is accomplished in TDMA-based technologies by enhanced data rate for GSM evolution (EDGE). EDGE replaces the modulation technique in GSM with 8-PSK (phase-shift keying), thus tripling the user's bit rate in that transmission technology's 200 kHz channels.

It is important to understand where, when and how these technologies will be deployed. Based on a currently available roadmap from various operators, Figure 2 illustrates the time-space evolution of 3G technologies.

Design and implementation challenges To understand the various challenges involved in design, implementation and maintenance of a 3G wireless network, one must understand the life cycle of a wireless network. The life cycle of a wireless network can be best explained as in Figure 3.

Wireless data applications have been defined by various standards' bodies. Separate migration strategies are proposed for each major air interface to complicate things further. When looking at the details, it is easy to lose sight of the basics. The truth is, even with the sophisticated technologies being deployed, the engineer's goal remains the same: The wireless data network design must provide coverage, quality, and capacity as defined by subscriber demand. Given that various migration strategies create slightly different problems, the design engineer needs a flexible way of modeling any data application to answer questions like:

- How many sites need to be deployed to support a specific wireless application ?

- Where should resources be deployed?

- How will the wireless application affect the QoS for voice traffic? How will capacity be affected?

- Does it make sense to deploy an intermediate technology, like GPRS or EDGE?

Next are some of the issues that the wireless community has to address before the next generation of wireless networks becomes a reality.

Mobile access speeds Qualcomm's High Data Rate (HDR) is a CDMA-based packet communication technology implemented on a separate network from existing voice services. In principle, HDR enables a maximum data rate of 2.4 Mb/s over a 1.25 MHz channel. However, these speeds are only possible at short ranges (less than 5% of a cell under ideal conditions), and averaged over a typical cell a theoretical maximum of 600 kb/s is more likely. These rates are, however, based on lab conditions and published expectations of average data data rates are closer to 200 kbp/s. As HDR is a CDMA technology, deployment may be limited to CDMA markets (United States, Japan, Korea), and even in the United States only PCS carriers can adopt HDR as 800 MHz operators are unlikely to be able to free up sufficient spectrum.

In the GSM and TDMA worlds, GPRS offers a theoretical maximum data rate of about 170 kb/s, although limitations in GPRS terminals will limit burst rates to 43 kb/s. Average throughputs will be even lower with published estimates of commercial services at only 28 to 43 kb/s. EDGE takes GPRS one step further, upgrading data rates to a theoretical maximum of 384 kbs, although average data rates are likely to be less than 128 kb/s. This represents a substantial improvement over GPRS and is adequate for some high-bandwidth business applications. However, the higher data rates increase modulation complexity, thereby decreasing the range of signals and necessitating significant investment by operators in a larger number of smaller cells.

3G Technologies like CDMA 2000 and W-CDMA promise theoretical burst rates of up to 2 Mb/s, capable of supporting multi-media applications such as mobile video conferencing. However, data rates of 2 Mb/s are only possible within the confines of a building and require a user to maintain a fixed position within 150 feet of the base station. Under such conditions, 3G wireless technologies are competing with wireless local area network (WLAN) technologies that provide data rates of 11Mb/s and do not have air time charges. Furthermore, available bandwidth will be shared and average rates per user will fall significantly.

Network coverage and quality of service The idea of wireless data represents a major evolution in current mobile communications toward a more global access concept with high data rate applications. The concept is to provide fixed, low mobility, and mobile wireless access to Internet sessions, video telephony, e-mail, file transfer, video conferencing and a host of other applications. Indeed, vendors are already trumpeting the capabilities of their handsets. This leaves most engineers wondering how to support such applications while maintaining voice capacity and quality.

As operators deploy data networks overlaid on existing voice networks, they will notice that presence of a mixture of high data rate handsets tends to raise the noise floor in the entire network, thereby opening up coverage holes. Figure 4 shows the simulated effect of 3G handsets (in yellow) interspersed with voice devices (in blue). Note the coverage hole opened as a result of the yellow handsets.

The quality of service (QoS) requirements of typical mobile packet data applications are diverse (e.g. consider real-time multimedia, Web browsing, and e-mail). Support of different QoS classes, which can be specified for each individual session, is therefore an important feature. Most 2.5G and 3G technologies (like GPRS) allow defining QoS profiles using the parameters service precedence, reliability, delay, and throughput. QoS issues include:

- The service precedence is the priority of a service in relation to another service. There are three levels of priority: high, normal, and low.

- The reliability indicates the transmission characteristics required by an application. Three reliability classes are defined, which guarantee certain maximum values for the probability of loss, duplication, mis-sequencing and corruption of packets.

- The delay parameters define maximum values for the mean delay and the 95 percentile delay within the network.

- Throughput specifies the peak bit rate and mean bit rate. So the QoS challenge for operators is well beyond what is in a voice-only network.

Deployment Costs The vast majority of traffic on current wireless networks is voice-based. Most wireless operators provide a limited (low throughput) data service, but usage is relatively small. Therefore, operators design their current networks based on a proposed demand for voice transmission alone.

The demand for wireless data services is predicted to be high and immediate. High demand will force the operators to take account of data transmissions when they design their networks. Immediate demand will involve the patching of current systems with data aware protocols.

To provide high data rate services over the wireless network, operators need to provide additional network investment. These investments are high in case of 2.5 and even higher for 3G.

Provision of a data service will have a direct impact on a network's existing cell capacity because there will be no increase in available bandwidth with data and voice transmissions coexisting (for current 2G networks). The predicted rapid uptake will produce an initial discontinuous jump in demand on an already (in some cases) stressed network. Such increases in demand will require considerable planning. Given limited investment resources, operators will have to find this extra capacity in their existing networks.

Data will be broadcast over a network as a bit-stream. Errors in this bit stream will require the system to perform a re-broadcast, resulting in longer transmission time and, in some cases, disputable, increased billable minutes of use. Dropouts will not be permitted. Interference, therefore, would directly affect the data transmission rate. Operators will control the quality of the data service by careful interference management. Using sophisticated RF planning tools and automatic frequency planning tools, operators can squeeze more capacity out of existing networks before investing heavily in infrastructure improvements.

Security With 3G networks proposing to provide seamless access to corporate networks, security of such an access mechanism automatically becomes an insomnia topic for the network operators, as well as for the information technology (IT) communities within the corporations. Digital wireless networks have a first level of data security built into them, which should suffice most casual application users (schedule and calendar management). Companies that plan on using these networks for sensitive data transaction must not rely solely on the measures provided by digital networks. Such companies should employ application layer data encryption technologies or public key/private key encryption technologies.

The downside of adding another level of encryption is that it affects throughput (E.g. CDPD networks have a theoretical data rate of 19.2kb/s). By the time you add a transmission control protocol (TCP) or user datagram protocol (UDP) for transport and your own data encryption, the data throughput is about 10 kb/s. Adding application-level encryption could cause throughput to decrease even further.

It's a billing thing Before operators can get their billing systems to bill data, several things must fall into place. The industry has to find a way to differentiate voice calls from data calls, then decide how to measure data. Once units of data are established, billing systems can be modified to accommodate them. Some mobile data systems today (like CDPD) have no way of distinguishing between call detail records (CDRs) of a voice call and a data call. So operators bill both calls identically (by minutes of use). With the advent of packet-switched services, users will be transmitting data by bytes and not minutes, and operators will tend to bill customers by number of bytes transmitted. Billing systems must be modified to measure packets.

Billing systems get their information from CDR records in switches. In a phased approach, CDRs will be modified first to include information about type of call (voice or data) and later include attributes of the call, such as number of bytes or packets transmitted and even QoS requested. Operators will also have to bill customers based on the QoS that user has requested (refer to the section on QoS).

Network scalability Scalability of solutions, as penetration and service usage increase, is another factor closely related to network coverage that needs to be considered in evaluating these technologies. Because the capacity of cellular networks is limited to the number of available channels at each cell site, as usage increases, average throughput will fall and service quality will deteriorate. When this happens, a greater number of access points need to be deployed and available spectrum needs to be increased. With 2.5G and 3G, the cost of increasing the number of access points increases linearly with each base station, offering fewer economies of scale. Given the high cost of each base station, this is likely to constrain network operators in their ability to scale networks rapidly and efficiently. Furthermore, while EDGE, CDMA 2000 and HDR technologies do improve the efficiency of spectrum use, the expected "crunch" in U.S. spectrum will make it difficult to increase network capacity to scale with demand.

Pricing policies Users are accustomed to paying flat-rate fees for wireline Internet access, in contrast to wireless voice services, which are charged on a per-minute basis. There are numerous problems with flat rate pricing in 2.5G and 3G. Both GPRS/EDGE and HDR are expected to be implemented alongside voice services with the result that data traffic has the potential to impinge on high-value voice circuits as more channels are used. This implies that data should be priced at a multiple of voice to account for the opportunity cost of capacity. However, based on initial access speeds, mobile data connections are likely to be lengthy, resulting in excessively costly service when charged at voice rates. Network operators will also face problems pricing at competitive flat rates due to the need to recoup the significant spectrum acquisition and R&D costs associated with implementing 3G networks.

Availability of "the Killer App" Wireless Application Protocol (WAP) is the technology available today to bring web to mobile phones (Japan's NTT DoCoMo uses iMode). However, there are several things wrong with this. Establishing a connection to the web takes up to a half minute and costs additional money (typically $10 over the monthly service charge). Accessing simple information like weather reports, stock quotes or flight information takes substantially longer over WAP than over traditional media or even the web. The display space for any information on a mobile phone screen is cramped at best (this is addressed by newer generation phones). The alphanumeric key pad is a clumsy input device; Finally, WAP designers need to recognize that telephones are designed for vocal exchange. In order for WAP to have wider acceptance, a voice-based interface is a must. All these obstacles could be overcome, if only consumers had better reasons to make a mobile phone connection to the Web. The industry needs a killer wireless web application.

Available tools Although the challenges associated with 2.5G and 3G are numerous and daunting, help is available to overcome them.

RF planning and design tools, combined with automatic frequency planning and cell planning tools, allow operators to optimize their existing infrastructure and performing "what-if" analysis involved with design of a 3G network. Network optimization and base station test tools allow engineers to troubleshoot individual base stations or optimize entire networks by driving the network to collect real data from live networks. Specialized network analysis tools allow pin-pointing of hand-off problems and the reasons for them.

Operators managing and operating a 3G network have a lot of complexity that they have to deal with. Various parts of these networks generate a tremendous amount of data that needs to be managed and translated to useful information. They cannot expect to succeed in their endeavors by using traditional point solutions. They need to have an operation system support strategy in place (OSS).

The bottom line As far as a network operator is concerned, the bottom line on 3G networks is based on three simple principles.

First, customers want new services to increase the productivity of their businesses and new applications to better compete in their market place.

The second is that new services require new networks. Traditional telephone networks spend 100 years doing one thing really well - handling voice traffic. Broadband revolution is just commencing and is here to stay. Older wireless networks were not built to handle the massive bandwidth required to support millions of users transmitting data. Over the past 20 years carriers have spent more than $1 trillion to support the steady but slow growth in voice traffic - they are expected to spend at least that much in the next 10 years to support cascading demand for next-generation voice and data services.

Finally, a carrier may elect to sit out the new investment cycle due to cash flow, capital cost or decreasing competitive position. However, its traditional services are becoming a commodity due to increasing competition. Therefore, although the need to invest may be unappealing, the alternative is slow erosion until death. Simply put, the choice is to chase the new world or die in the old.

The question is not whether mobile data technologies are becoming available, but whether operators and network equipment manufacturers can build and sustain the networks that will be able to satisfy the demands of a data-hungry, "always-on" society.