Quantization in Task-Driven Sensing and Distributed Processing

Quantization is the mapping of continuous quantities into discrete quantities, an operation far more general and flexible than the ubiquitous example of analog-to-digital conversion of scalar amplitude values. By appropriate choice of distortion measures and transmission constraints, quantization can incorporate signal processing such as statistical classification, estimation, and modeling. We here survey several approaches to incorporating such tasks into the quantization and possible extensions to distributed signal processing