Custom designed DSPs will be a better hardware platform for software defined radio (SDR) in mobile phones than reconfigurable logic, according to Professor Christer Svensson, of Linkoping University in Sweden.
“Certainly we need SDR for next generation handsets and other equipment, and we need it badly. But, configurable logic (as we know it, that is FPGAs) is certainly not the solution,” Svensson told EW. “FPGA uses 35 times the area and 14 times the power of an Asic, which is not affordable in portable equipment. The solution is instead application specific DSPs, developed by firms such as NXP, Icera and Coresonic.”
A DSP device, for example, can use a quarter the area and the same power as an Asic. The device is fully programmable. Svensson is also CEO of Coresonic.
But for the FPGA firms, SDR looks like a great opportunity. “It’s the application we’ve been waiting for. We never had an application until software defined radio came out. SDR is the killer application of reconfigurable logic,” Wim Roelandts, CEO of Xilinx, told EW, adding, “now that we know the application we can supply the tools.”
Whereas NXP, Icera and Coresonic are working on SDR basebands, there is also the issue of how to put multiple RF onto a low-power chip suitable for a handset.
“We’ve a long way to go before there’s real convergence on the handset. We have different chips for WiFi, NFC, UWB, Bluetooth, FM radio and so on. We have to converge the RF, but we can’t put seven radios on one chip,” said Dr Henry Samueli, CTO of Broadcom, “so radio has to reconfigure itself. The same piece of silicon has to be able to reconfigure itself as a Bluetooth radio, or a UWB radio, or a WiFi radio, or an NFC radio.”
Svensson is working on an approach to this problem at Linkoping University.