Signal Processing with Teams of Embedded Workhorse Processors

Advanced signal processing for voice and data in wired or wireless environments can require massive computational power. Dueto the complexity and continuing evolution of such systems, it is desirable to maintain as much software controllability in the fieldas possible. Time to market can also be improved by reducing the amount of hardware design. This paper describes an architecturebased on clusters of embedded “workhorse” processors which can be dynamically harnessed in real time to support a wide rangeof computational tasks. Low-power processors and memory are important ingredients in such a highly parallel environment.