Compressive quantization in software defined receivers

Modern analog-to-digital converters (A/Ds) are composite structures that contain one or more internal A/Ds. The composite A/Ds carry out significant amount of mixed-signal processing in addition to the quantization performed by the internal A/Ds. Exploiting specific features of mixed-signal processing and/or specific properties of digitized signals enables improvement in performance of the composite A/Ds without increasing the burden imposed on their internal A/Ds. In this paper, this approach to the A/D development is named ¿compressive quantization¿ by analogy with ¿compressive sampling¿ that allows sampling rate reduction for sparse signals below the Nyquist rate. The notion of compressive quantization is intended to bring together various techniques and methods that improve performance of the composite A/Ds not by "brute force" (i.e. by increasing the number of bits and speed of the internal A/Ds), but by more intelligent processing of signals in composite A/Ds. It provides a new viewpoint intended to improve effectiveness and efficiency of quantization. To that end, this paper systematizes various quantization methods within the compressive quantization framework. Several methods of compressive quantization that utilize specific features of mixed-signal processing in A/Ds and several methods that utilize specific properties of digitized signals are considered. Feasibility and benefits of these methods are discussed. The paper also demonstrates the diversity of possible approaches to compressive quantization in software defined receivers (SDRs).