A methodical approach for stream-oriented configurable signal processing

The characteristics of the signal processing tasks associated with an advanced wireless receiver are well matched to the capabilities offered by CCM (custom computing machine) technology. Collectively, digital receiver algorithms seem to share the following properties: (a) repetitive operations are performed on huge data sets, (b) the dominant computations are conducive to very deep computational pipelines, (c) a moderate amount of latency can be tolerated, and (d) different environmental conditions require different signal processing methods, which in turn require distinct computational structures (time-varying computation). In addition, when coupled with run-time reconfiguration, the amalgamation of signal processing tasks may be compactly combined in a power-efficient computing module. This paper focuses on the design methodology for implementing large and intricate stream-oriented signal processing tasks.