Wireless cellular services originally were driven by the need for wireless speech communication. However, we are seeing more data-based services taking a major role in our lives-paging, e-mail, Internet access, electronic commerce and others. Although some services traditionally have been offered only over wireline infrastructure, the high mobility of our society requires more and more wireless access to these services.
Up to now the barrier has been the limited bandwidth of the wireless systems. To enable wireless access to these emerging services, wireless media must be able to provide a cost-effective wireless "pipe." Enter 2.5G and 3G, which are all about providing this pipe.
Those new standards have three main attributes. First is higher data rates. Essentially, 2G systems' data rates are limited by the bit rate offered by an active compressed speech channel. This yields data rates of about 14.4 kbit/second, hardly enough for these services. But 2.5G and 3G systems are targeting data rates of up to 384 kbit/s and, later, 2 Mbit/s. This requires not only changes in the air interface, but also in providing the ability to dynamically allocate multiple slots to a user that requires higher data rates.
The second attribute of the new standards is packet switching, a technique that partitions digitized voice and data into packets that can be loaded conveyor-style with other packets onto the airways. It is replacing circuit-switched techniques in which a connection between transmitter and receiver is maintained even when nothing is being sent. While today's wireless system are circuit switched-for example, a connection is established and main-
tained, no matter how much information is actually transferred-the next-generation system will be packet switched. That is, information is transferred as packets. Packet switching is better suited for data transfers and provides much better bandwidth utilization between the base-stations and the network.
The third attribute is the voice side, where there is a push toward higher speech quality. Wireless speech quality has always been inferior to that of wireline. The reason is the three-way trade-off among speech quality, compression ratio and bit rate, which translates into system capacity and implementation complexity, which translates into cost. Using the sharp cost-reduction curve in DSP technology, next-generation systems will use higher-quality vocoders, which do not increase the bit rate.
How will the wireless topology be affected by the introduction of 2.5G and 3G systems? Most 2G systems are based on a four-level hierarchy. At the cell center we see the base transceiver station (BTS), which supports a small number of radio links to the mobile units traveling within the cell. The base station controller (BSC) is responsible for the handoff procedure when a mobile unit moves from one cell to another. The Mobile Switching Center (MSC) provides the switching function in the system. It is the gateway to the public-switched telephone network and records billing and roaming information. Between the BSC and the MSC we find the transcoder function, which provides speech compression/decompression as well as echo cancellation to compensate for the delay in the system.
This topology will not necessarily remain intact for next-generation wireless systems. Packet switching opens the door for links into packetized networks, such as IP, right from the BSC level, thus bypassing the traditional wireless switch. We might very well see multiservice systems that ignore the traditional boundaries between wireless and wireline.
Meanwhile, cost is another driving factor. Driven by competition and deregulation, wireless service is becoming less and less expensive. In the wireless infrastructure equipment cost is driven not only by the price of silicon components but also by space and power dissipation. This in turn translates into much higher integration, lower power devices and much denser, less expensive packaging. This, in fact, is fueling a trend toward the use of LDMOS for the power transistors in basestation transmitters, replacing costly GaAs devices.
No single standard
The migration from analog to digital systems did not meld the wireless market into one worldwide digital standard. It seems this will also be the case in 2.5G and 3G. Wireless infrastructure manufacturers that support multiple regions have to design different equipment for different standards. Significant cost reductions can be achieved by utilizing the same hardware platforms using different software to support different standards. Therefore, a trend in which more functions will be done in hardware than software will likely materialize. But to minimize software development time, thereby improving time-to-market, the DSPs used must enable efficient software development in C. This, in turn, is spurring the trend toward C compiler-friendly DSP architectures.
The DSPs are used for voice and channel coding. Very long instruction word (VLIW) architectures, like the StarCore SC100, become efficient targets for C compilers. This will enable software programmers to complete their work in C. Ordinarily, DSP programming requires assembly language experience-along with an intimate knowledge of the internal register and pipeline structures of the DSP. With new-generation DSPs, tuned to compilers, the C compiler automatically accounts for the hardware resources of the target DSP. Thus, VLIW architectures will extract high performance from programs written in C. The use of C language programming will enable faster time-to-market, and easy upgradability for cellular basestations in their move to 2.5 and 3G capability.