Mixed-Domain Systems and Signal Processing Based on Input Decomposition

Abstract

Using input decomposition as a starting point, a variety of new types of systems and signal processors, which mix together domains traditionally kept separate, are derived, and their properties are examined. In one of these systems, digital signals are processed in continuous time, thus avoiding sampling and consequent aliasing, while maintaining the advantages of digital implementations in terms of programmability and noise immunity. In another approach, digital bit waveforms are processed directly by conventional analog filters. In a third approach, a nonlinear input decomposition results in systems in which signals are processed with a compressed dynamic range; this is done in a way that avoids output transients that characterize other approaches to compression, and also avoids the need for precise nonlinearity control. Several other possibilities are discussed. In all cases, the resulting systems can be input-output linear