Kaivan Karimi investigates the hurdles which still have to be cleared before the availability of services such as video, internet/intranet access and multimedia messaging become widespread.

The transition from 2G to 3G mobile standards was sparked by the promise of high-speed data services. 3G systems support data rates of up to 384kbit/s, and eventually even 2Mbit/s, which is around 100 times higher than those provided by existing 2G technologies. Services such as video, internet/intranet access and multimedia messaging are now possible.
The limiting factor of any radio communication system is the transmission medium itself. The information bearing radio waves are susceptible to many degradations as they propagate from a base station to the mobile terminals travelling through a cell. Electromagnetic waves emanating from a base station antenna are reflected and diffracted by large obstacles, causing scattering and dispersion of the signal. This implies that multiple copies of the original transmission arrive at the receiver with different delays and from various directions, and also that they interfere with one another either constructively or destructively, leading to signal fading. This multi-path propagation phenomenon is typically not static in a mobile environment, hence giving rise to a time-varying radio channel characteristic which imposes a number of challenges on the receiver, especially for high data rate transmissions. In 2G systems equalisers are used to combat the intersymbol interference caused by multi-path propagation. In 3G systems based on WCDMA (wideband code division multiple access) multi-path components that are sufficiently separated in time can be resolved and combined in a signal reinforcing way by an appropriate 'rake' receiver.
While the 'data hype' marketing may have made people expect that once 3G services roll out everyone in a cell site can use high-speed data services, such as video-telephony, the reality of the matter looks somewhat different. Each cell can only sustain approximately five high-speed (384kbit/s) data users, and in fact the presence of even a single high-speed data user in a cell can hamper the transmissions of concurrent voice calls.
Furthermore, since the quality of service (QoS) is largely a function of the distance from the base station, the highest 3G data rates are really only available in the central 10percent of the cell area nearest the base station. This is the 'hidden truth' behind using high-speed data in CDMA systems in general, especially WCDMA systems that offer UMTS/3G services. Add to all this the higher occurrence of blind spots and limited in-building penetration, and the ability of operators to provide the promised coverage, capacity and QoS seems quite doubtful.
Today, after months of delay in 3G network rollouts, the issues related to QoS, capacity, and coverage may seem secondary, and taking a back seat to the roll out of basic 3G services. However, if the experience of Japanese 3G network rollouts, which are the trailblazers in this domain, is any indication, then very soon after the basic rollout the realities of high-speed data services will become apparent: while you can maintain a high-speed link when you are next to the base station, as soon as you drive five blocks away you can no longer do that and have to switch from video to email.
However, all is not doom and gloom, and it is worth examining some of the new technologies and emerging concepts that can address this situation. Various methods for dealing with multi-path fading already exist, for example. The next step is to use and, in fact, cause the occurrence of multi-paths to boost the quality of the signal. This is what diversity processing is all about. The concept is simple: equip the terminal with the ability to make sense of a large number of signals from two antennas, resulting in an improved signal and an increased signal-to-noise ratio at the mobile terminal. The benefits are obvious - data services can be offered in more of the cell area, fewer calls will be dropped, in-building coverage will improve, operators will need to spend less on infrastructure, and network capacity will increase.
Diversity processing is defined as taking multiple paths between the transmitter and receiver and combining them in order to maximise the received energy. To do this, a number of transmission paths are required, all carrying the same message but with independent fading statistics.
For interpreting signals that exhibit spatial diversity, one transmit antenna and x receive antennas are required. The receive antennas see different and independent signals, provided they are sufficiently far apart. The minimum separation necessary is dependent on the environment, but half to a third of a wavelength separation is often enough at the mobile terminal to achieve a substantial antenna diversity gain. The concept also holds for one signal-receiving antenna and many transmitting antennas. ie systems that employ base station transmit diversity.
Temporal diversity is employed to combat the effect of the mobile terminal going through fades. It can be caused intentionally to aid reception. If information is sent repeatedly the chance of the terminal being in a fade when the signal is received multiple times is reduced. To achieve this effect, the separation time of the signals must be 1/fD where fD is the Doppler frequency. Interleaving has the same effect.
Signals transmitted on two orthogonal polarisations in the mobile radio environment exhibit uncorrelated fading statistics. This is called polarisation diversity.
Finally, frequency diversity involves the transmission of a narrowband signal over multiple frequencies, where the frequency separation must be greater than the coherence bandwidth of the channel. The coherence bandwidth represents the frequency separation of uncorrelated signals. Spread spectrum systems such as WCDMA inherently implement this kind of diversity, since the signal spans multiple coherence bandwidths.
Once multiple independent paths have been created there are two ways of combining them to improve the communications link. Switch diversity is the simplest. For narrowband signals the terminal is instructed to use antenna branch number one until the signal power falls below a certain threshold Y, then switch to branch two. The terminal is instructed to use branch two until the signal power falls below the same threshold Y, then switch to branch one again. For wideband signals a different criterion may be better, for example using a measure of inter-symbol interference or a combined measure of ISI and signal power.
The second way of combining the independent diversity paths is selection diversity. This involves making measurements of the received signal power on both antennas, and then selecting the one with the larger signal power for data reception. In a Rayleigh fading channel for 1 per cent outage, this delivers a 10dB saving in the link budget for two branches, or a 16dB saving for four branches. Maximal ratio combining measures the signal amplitude and phase on each antenna, scales the output of each according to its SNR, co-phases the signals and sums up the scaled results. Alternatively, equal gain combining just co-phases the signals and adds them together. This list of methods to improve signal and coverage is by no means exhaustive and various baseband solution manufacturers have come up with their own variations. Innovics Wireless' IW7500 processor, for example, performs 2D combining to deliver optimum performance in conjunction with the rake receiver. Their particular algorithm dynamically trades off diversity, beamforming and spatial interference nulling based on the channel characteristics, resulting in a 7dB average SNR improvement in WCDMA systems.
The challenges facing diversity processing implementation are mostly concerned with power consumption. This needs to be at a minimum so that talk time is not sacrificed for better capacity, coverage and QoS. Recent developments in VLSI silicon technology and new algorithms and antenna designs mean that diversity processing can be carried out in the terminal as well as the base station.
Detailed calculations show that diversity processing can actually be used to reduce power consumption in 3G systems. The baseband processor can be designed so that if the receiver is within the 10percent of the cell area supporting the highest data rates, the diversity processing elements are switched off. Elsewhere in the cell, a diversity-equipped terminal would be able to maintain the highest possible data rate, albeit with a slightly higher power consumption, for example, approximately 15percent in Innovics' case. However, the drastically increased data rate means that the terminal might typically take 85percent less time to complete downloading a large file. This decrease in download time means that terminals expend just 20percent of the energy required by a conventional solution. In addition to this, one additional RF chain for a second antenna requires an 8percent increase in power, but delivers a 500percent improvement in average throughput rate.
Soft and Softer hand-off techniques are a part of any CDMA (and hence WCDMA) system. This is an area where the mobile user interacts with the other sectors of the same cell site, or other cell sites, and hence the improvements in this interaction have 'system wide' effects. An area that requires serious consideration is how to reduce the hand-off 'zone', and in effect not only optimise the power consumption of the user equipment but also reduce the utilisation of the resources of the entire system, including the neighbouring cell sites, and thus increase the capacity of the system. The faster the communications with other base stations happens, and the faster the hand-off takes place, the smaller the hand-off zone. This can be achieved by adding extra rake fingers that are dedicated to expediting this process.

The story does not end here, as 3G is much more complex than 2G. Demands have increased along with the complexity of the computational requirements. For example, processing 20 multi-paths requires a computational horsepower on the order of 6GOPS (six thousand million operations per second). Fortunately, new design and manufacturing techniques such as system-on-a-chip (SoC) and system-in-a-package (SiP) mean that designers can place microprocessor cores on the same device as hardwired implementations of algorithms, memory and peripherals. Tasks can be partitioned between hardware and software implementation and both parts can be designed simultaneously. Designs can be optimised for size, power consumption and cost. However, designers must still take into account the fact that 3G is still fluid.

Kaivan Karimi is Director of Business Development, with Innovics Wireless.