Rethinking of Paradigms in Signal Acquisition

The aim is to rethink the approaches in signal acquisition. Classical paradigms often lead to highly redundant and inflexible measurement signals with additional subsequent steps for evaluation. Redundancy involves additional expenditure while in most cases, only few parts of the massive amount of data is expedient. This outdated approach is expected to be replaced by the new philosophy in acquisition, to focus on relevant data already in the acquisition step and to exploit accessible prior knowledge. This allows the development of adaptive, self-learning sensors technologies which derive the next appropriate measurement from the previous acquired data and the constantly changing test environment, and which monitor themselves.

Self-learning Sensors Technology for Optimal Measurement Planning

Complex sensing systems, in particular those focused on physically interpreted signal generation and interaction volumes applying acoustical, ultrasound-based, thermal or radiographical techniques for materials and products and using sophisticated and closely interacting microelectronics, generate a massive amount of data (big sensor data with several terabytes). The acquisition of this data and its processing to relevant signal information requires short processing times near real-time. The contained information is supposed to be compressed without loss, transferred in following data streams and allocated for further processing.
Such an approach demands measurements and subsequent signal processing not to be considered separately but as one unit. It is feasible when appropriate, physically motivated models of the measurement process are accessible which allow derivations for optimized measurement planning. Therefore, the modelling of test processes is essential for this project.

The aim is to establish algorithmic developments and microelectronic implementations for ‘slim’ (narrowed down to the most necessary) signal generation, and techniques for adaptive signal acquisition and electronics-related, lossless signal processing reduced to relevant data.