Wideband code-division multiple-access (W-CDMA) is emerging as one of the main technologies for the implementation of third-generation (3G) cellular systems. The complexity of W-CDMA systems can be viewed from different angles: the complexity of each single algorithm, the complexity of the overall system and the sheer computational complexity of a receiver.
Implementing a W-CDMA receiver requires an understanding of the algorithms that convey information between transmitters and receivers in a W-CDMA system. From a design-tool perspective, users need to develop or otherwise access advanced coding and transmission algorithms. They need model aspects of complete systems, such as the transmitter, propagation channels, RF front-end, initial synchronization between base station and mobile phone, and their tracking. The W-CDMA handset is essentially a rake receiver, which attempts to concentrate RF spectral energy that has been spread on several axes. An additional design aid is the ability to perform fast simulations. The rule of thumb is that W-CDMA link-level simulations are over 10 times more compute-intensive than current second-generation simulations.
Standards for second generation (2G) systems include GSM, PDC, IS-136 and IS-95. GSM is used in more than 100 countries and accounts for the majority of the world's digital cellular users. The Personal Digital Cellular (PDC) system enjoys success in Japan, where the demand is about to exceed the spectral resources.
Basically, 2G systems were introduced for one main application: voice communication. Improvements sought to achieve a better spectral efficiency than traditional analog systems. That allowed service providers to accommodate more users per Hz. Recent developments are meant to enable better data transmission over the same interface, especially packet data and higher data rates.
Higher data rates are the main requirements for 3G systems and they include, in particular, support for a wide range of data rates, from typically a couple of kbits/second up to several Mbits/s for a 5-MHz bandwidth. Different users in the system can simultaneously transmit at different data rates and those data rates can even vary in time.
With the W-CDMA Layer 1 proposal, two 60-MHz paired bands are used for the frequency-division duplex mode-at least for 3G systems in Europe and Japan. The "uplink" uses the lower 60 MHz and the "downlink" the higher 60 MHz. Additional frequency (unpaired spectrum) can be gained in the time-division duplex mode; here, uplink and downlink share the same carrier frequency but alternate in time.
Essentially, the current W-CDMA proposal defines two types of transport channels: common and dedicated. Common channels in the W-CDMA specification include broadcast, paging, forward access (FACH) and random access (RACH). Those channels transmit information meant for several users on the downlink, or the information can be shared by several users-with a risk of collision on the uplink.
Dedicated channels (DCH) carry information that is specific to a particular mobile phone or mobile base-station link. Voice information from a phone call is typically carried by such a channel. Dedicated channels use a fast power-control scheme-an update frequency of 1,600 Hz-to mitigate as much as possible the influence of the near-far problem on the reception of the CDMA signal.
A W-CDMA handset must perform several functions that use those channels. The functions include initial synchronization such as fast cell search, channel estimation and rake combining, power control, antenna arrays and multiuser detection.
When a mobile phone is switched on, it has to acquire the fundamental parameters of the system, which can be summarized by base-station identification. That involves determining which base station is the best one to log on to, its transmission frequency, its timing and the scrambling code. Since the signal sent by a base station has been scrambled with one of 512 possible scrambling codes, the mobile phone cannot directly listen to the common control channel. To allow the mobile phone to reach speed more quickly, every base station continuously transmits an additional piece of information called synchronization code or search code. Because this code is not scrambled, it is not orthogonal to the other signals but can be readily detected without deciphering the scrambling code or its timing.
The mobile phone first looks for that information, which is sent at the beginning of each slot. Through correlation, the mobile phone can determine the position of the synchronization code in the signal, which also indicates the beginning of a slot. With more information-a second synchronization code-the mobile phone can determine the frame timing and identify a subset of 16 possible scrambling codes. In a last step, it simply tries out all 16 possible codes by attempting to decode the common control channel.
At that point, the phone has performed an initial synchronization and can listen to the broadcast channel, which informs it about system and cell-specific information (like the allowed codes for RACH bursts). Once this information has been acquired, the mobile can RACH burst a request to the base station for resources.
Under ordinary circumstances, the wireless propagation channel impairs the received signal. The propagation channel can be characterized mathematically by several independently fading Rayleigh components. Because of its very structure (CDMA), the signal tends to be orthogonal to a shifted version of itself, especially with large spreading factors. In a typical approach, a receiver can be built that combines the energy of all the echoes. This structure is called a rake receiver.
The rake receiver is composed of several identical processing units called rake fingers. Each rake finger is assigned to demodulate a particular echo. One finger synchronizes itself to a particular echo and performs joint descrambling and despreading. The estimates at the output of all the fingers then have to be combined.
The optimum combining method, assuming a white Gaussian noise-that is, the one providing the best signal-to-noise ratio at the rake receiver output-is the maximum-ratio combiner (MRC). This combiner cophases all the estimates at the finger output and multiplies them by the amplitude of the echo for each particular echo term.
To perform MRC combining, the different echoes have to be identified, particularly their position and complex amplitude. This is the task of a channel-estimation algorithm. A classical way to perform channel estimation is to rely on the knowledge of a bit sequence transmitted along with the useful signal, and perform cross-correlation between the expected bit sequence and the received signal such as a cyclic redundancy check.
In the W-CDMA standard, pilot symbols are transmitted as part of the dedicated physical control channel. Transmit power control (TPC) bits are also transmitted for the closed power-control loop and an optional transport format combination Indicator field gives information on the transmission rate of the current frame.
Since the spreading and scrambling codes are known, the pilot symbols translate into a known bit sequence. The auto-correlation properties of this bit sequence allow the receiver to assume that a peak in the cross-correlation with the received signal corresponds to the location of an echo. The channel estimator can then select the best echoes and pass the timing and complex amplitude information to the rake receiver.
Power control is a key issue in the specification and design of any W-CDMA system. The current W-CDMA proposal specifies the transmission of TPC bits in each time slot of a dedicated channel. This translates into a power-control frequency of 1,600 Hz.
These TPC bits indicate whether the transmit power should be reduced or increased by, for example, 1 dB in the next slot. It is important to jointly optimize the slot structure and the timing relationship between uplink and downlink slots to minimize the latency of the power-control loop.
The remaining issue is how the mobile phone and the basestation know how to set the TPC bits. The power-control scheme is signal-to-noise plus interference ratio (SNIR)-based. The receiving unit thus has to estimate the current SNIR and compare it with a threshold, provided by the network. If the SNIR is greater (lower) than the threshold, a power down (up) command is sent back to the transmitting device. SNIR estimation can be performed with the help of the pilot symbols.
Moreover, the existence of dedicated channels and the presence on each of these channels of pilot bits enables the use of directive antenna using beam-forming techniques. The main advantage of beam-forming is that it reduces the interference level within a cell, allowing an increased number of users per cell. The dedicated pilot bits enable channel estimation for the particular beam-formed connection.
Significant efforts are under way to find the right algorithms and the corresponding architecture. This task calls for a well-organized design flow that is flexible enough to handle both the system and algorithm aspects as well as the path to implementation through behavioral-level or RTL synthesis. This is one of the major issues to tackle in order to have a product early on.
Synopsys' approach to these issues is to provide a flexible combination of tools, environments, advanced algorithms and services to help streamline or even jump-start the design process. It is also crucial from a methodology point of view that a company use a consistent platform for development, since many of these simulations are performed by large groups of engineers, often located at different sites all over the world.