When compared with second-generation mobile systems, it should be realized that the communication processes for 3G are perhaps 100 times more complex before the application load is considered. A DSP-only solution will quickly run aground and require hardware accelerators to process the tasks that the DSP simply cannot do in time.

A second way to approach the design problem is to develop hardware functions from scratch that can process the 3G baseband data without the need for a DSP. This undoubtedly will entail a longer development process until functionality is complete, but it has the clear advantage in terms of design structure and labor division. It is relatively straightforward to allocate tasks as independent functions to a team of engineers.

One downside is that a larger area of silicon is required to produce the same functionality. Needless to say, that silicon area has to be balanced against the area of silicon required to store the software for a DSP.

The hardware constructions are each dedicated to a particular process and can reduce power needs by a factor of a hundred or more compared with a DSP.

Of course, they are also much less flexible than a DSP, but since they are only designed for one task, that is not a problem. The main argument against their use is that any change in the product specification would require a new design. For that reason, sufficient modification capacity is included in the design.

This usually only increases the size by less than 10 percent, and when a systems designer is involved in the process, the resultant product will last a long time.

The product made with a DSP will be one that uses more power than expected, while the design of the second method will be larger than expected.

Alex Lax is Chief Technology Officer of Cellular 3G (Lexington, Mass.).