Last week, scientists came together at the Conference On Telecommunications and Computers in Portsmouth, England to discuss software-defined radio (SDR). As Francis Kinsella, an engineer at EADS Astrium sees it SDR is still an emerging technology. It is Kinsella's view that two main obstacles that have held SDR back are the speed at which analog-to-digital converters transform radio signals into a digital datastream and the available computing power.

"But we foresee advances in both of those sectors that could really mean an explosion over the next five to 10 years for SDR," he said.

In fact, virtually all the researchers at the conference agreed that SDR equipment will soon become commonplace.

The aim for SDR, in both receive and transmit, is to be able to translate and process virtually any mode or frequency of radio signal. Another way to look at SDR is this: If you can configure hardware that will accommodate infinitely wide receive and transmit paths, fill it with memory and omnipotent DSP capability, then you can handle all manner of upconverting, downconverting and modulation schemes — by simply changing the software. Of course, that is a tall order, but it is surprising how close recent innovations have already brought SDR to it ultimate objectives.

Commenting on the remarks made at the Portsmouth conference, Paul Mesibov, director of engineering, at Pentek Inc., a U.S. supplier of preconfigured software radio subsystems said, "The key to bringing SDR from a dream to reality is the availability of high-performance, programmable platforms that include all of the I/O and processing resources that can be provided in a highly integrated manner."

As Mesibov sees it the crux of the development issue is the ability to enable waveform developers — who are the people who actually write processing code — on whatever platform they see as appropriate. They can then respond to the kinds of modulation schemes — in other words, waveforms — that someone is interested in. So if you are developing a battlefield radio that is to employ many different communications schemes — you need to have all those schemes programmed into it. That is why developers must have high-performance computing platforms with intimately connected I/O and FPGA processing capability.