Academic Seminar (Course)
FENS G035
08.05.2024 - All Day
SEMINAR:Ultra-low power and robust analog computing for biosignal monito
Improvements in physiological sensing and processing have facilitated the timely diagnosis of disorders through wearable and implantable systems. However, the growing volume of high-resolution biosignals necessitates more efficient sensing and processing systems, especially in environments constrained by power and space. This is particularly critical in the development of unobtrusive wearable or implantable devices, where longevity and miniaturization of batteries are key objectives. Traditional wearable or implantable biological signal sensing systems typically consist of an analog-front-end for signal-to-noise ratio enhancement, an analog-to-digital converter for digitization, and a radio for transmitting the digitized signals. The radio component, being the most power-intensive, poses a significant challenge, particularly as data bandwidth increases. To address this, extracting physiologically-relevant information at the sensing level can substantially reduce data bandwidth. This talk will present two approaches to achieve ultra-low power and high accuracy on-chip processing in resource-constrained environments. Firstly, it focuses on implementing a high-accuracy digital biological signal processing algorithm within the analog signal processing domain. This ASP implementation demonstrates superior electrocardiogram (ECG) feature detection with minimal power consumption, a notable achievement. Secondly, a novel biosignal processing algorithm grounded in physical principles is introduced for detecting intracortical neural spikes and ECG features. Furthermore, an ultra-low power physical implementation of this algorithm in silicon will be presented.