Information-based Analysis of Compressive Data Acquisition

Data acquisition in compressive sensing (CS) is commonly believed to be less complicated, and even less costly, while performing agreeably. However, this optimistic belief in the performance and complexity of a CS sensor is still to be quantified. We analyze the performance of diverse compressive data acquisition schemes with the same input signals starting with spatial measurements from antenna-array processing. The performance (given in the familiar terms of processing gain, detection, accuracy and resolution) is elaborated by using the framework of information distances from information geometry. Only with a scheme of compressive data acquisition starting directly at reception, i.e. at antenna (with no receiver noise from the low-noise amplifier yet), CS could be simpler and still perform as good as, if not better than, existing sensing. Major requirements of this coded-aperture scheme as well as possible implementation by using metasurfaces are indicated.