Complex system calibration through sampling of constitutive structures: Examples from circuits and codes
 
In a wide variety of complex systems consisting of many components each carrying some randomness –be it a  code as a collection of individual symbols,  a hardware memory consisting of many  infinitesimal circuit cells, or a wireless network comprised of many mobile users – the exact analytical evaluation of the performance often becomes  intractable  or inaccurate. In order to guide future innovations in system design, it is of crucial importance to develop approaches to efficiently evaluate the performance of a complex system, even though the lack of an explicit method makes this problem very difficult in general. We propose a statistical sampling based viewpoint which (1) identifies constitutive structures in a complex system, and (2) efficiently exploits these structures in a suitably developed statistical sampling algorithm. Results from circuit and coding domains demonstrate the effectiveness and the application range of the developed methodology.