Digital Signal Processing – Going Embedded, Real-time, and Multi-Core

Digital Signal Processing impacts numerous industries and applications including audio, video, consumer electronics, multimedia, speech recognition, biological systems, wireless, telephony, broadband communications, and more. Application examples abound from ADI, TI, and Xilinx to name a few. Xilinx teamed up with Pixel Technology to perform 3-D facial recognition technology with 1000x performance over traditional processors by using a stacked architecture of FPGA and DSP processor units. TI is targeting WiMAX with their DSP platform. Chipwrights uses a vector architecture to process several data streams at once for image processing on hand-held devices.

One of the main trends in DSP is the shift to embedded applications. General purpose DSP is dominated by the cellular industry with China currently driving the demand. Embedded DSP is driven by the consumer electronics industry and has a broader reach with about 2x the volume. The DSP-FPGA.COM site is a good source of information and has a news release on the topic.

Another trend is the shift from single to multi-core processors. This article describes how this shift is happening not only in microprocessors but is a natural fit for DSPs as well.

Other trends include DSPs working with real-time systems. This article describes several strategies for handling real-time data flows including polling a status bit to see if new data has arrived and using interrupts and stacking up the I/O for follow on processing.

If you are unfamiliar with DSPs, you can find a tutorial on it, here.

Two resources for more information are the Global Signal Processing Times and DSP-FPGA.COM

Best regards,
Hall T.