B2.1 - Robust & Technology Agnostic Integrated Implementation of Future Sensory Electronics based on Spiking Information Processing for Industry 4.0
- Event
- AMA Conferences 2017
2017-05-30 - 2017-06-01
Nürnberg, Germany - Band
- Proceedings Sensor 2017
- Chapter
- B2 - Sensors for Industrial Applications
- Author(s)
- A. König, A. Chandra Kammara - Technische Universität Kaiserslautern (Germany)
- Pages
- 177 - 182
- DOI
- 10.5162/sensor2017/B2.1
- ISBN
- 978-3-9816876-4-4
- Price
- free
Abstract
The ongoing advance in integrated electronics and sensors technology jointly with recent application challenges, e.g., from automotive, home automation to Industry 4.0, imposes stringent requirements on sensor conditioning and conversion electronics with regard to diversity, genericness, accuracy, and robustness hard to be met by traditional amplitude coding. Moving from amplitude to time domain is a valid alternative. In our work, we pursue a spike-coded approach based on peripheral neural assemblies. Bio-inspiration, e.g., Tropotaxis based methods, the Jeffress’ model of sound localization and snake prey detection using its tongue are employed in a spiking architecture to avoid the sensitive scaling networks of the common approaches. Two corresponding architectures have been evolved, designed, and the first one has been advanced to the manufacturing of a first 350 nm CMOS chip, amenable, e.g., for XMR sensors, with variable bit resolution. First test results in a prototype system show principle expected functionality, but also the need for continuous inherent adaptation, which will be pursued in future work in the next generation of robust spiking sensor electronics chip. The second approach has been simulated as a proof of principle design creating a 16 bit ADC with a sampling rate of 50 kHz, with a design which can be rapidly scaled improving all its parameters.