Wiki source for PhdThesisAbstract
=====Energy-Aware Signal Processing for Wearable Biomedical Sensor Systems=====
""PhD"" Thesis / Doktorarbeit
Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), Technische Fakultät
2024
Autor: Andreas Tobola
====Abstract====
Wearable technology has been rapidly changing the overall picture of medical diagnosis. Highly integrated hardware and smart algorithms are enabling long-desired healthcare applications. Nevertheless, engineers still face major challenges. One such bottleneck is the limited energy of sensors worn on the body. Despite highly integrated hardware, the energy limitation requires larger batteries and eventually energy harvesting in addition.
This thesis gives a comprehensive insight into development of portable sensor technology, especially for biomedical applications, with a particular focus on minimizing energy consumption. One of the central drivers of this work is the philosophy of taking a holistic view of the entire signal processing chain, from the sensor to the evaluation of biomedical information. For this purpose, all parts of a biomedical signal processing chain were examined. The examination included hardware and software components.
Specialized tools and methods were developed for this purpose. A new power profiling method was developed to evaluate the power consumption of microcontroller-based systems. The unique feature of this method is the automated profiling tool which analyzes the system’s power consumption for a better understanding of the underlying processes. A further contribution of this thesis is new sensor hardware and new software. This sensor system was particularly designed for low-power evaluation. The sensor system simplifies low-power debugging with modularity and the comfort of dedicated power measurement terminals. Different variants of the sensor system hardware and software were developed and evaluated. The hardware variants investigated in this thesis were various sensor circuits, power supply circuits, and microcontroller types. Examined software variants were real-time operating systems, software-based power managers, and signal processing algorithms. The outstanding feature of the system is its scalability, achieved by designing software and hardware components configurable. The configuration determines the behavior of the sensor system. Also, the configuration influences power consumption. For the Ultra-low-power Sensor Evaluation Kit (ULPSEK), the average power consumption was determined between 50 µW and 10 mW, depending on the configuration.
The measurement results from the test bench, combined with theoretical principles, were merged into a general mathematical energy model to support the engineer in planning a new sensor system. Compared to state of the art, the model combines two standard sleep modes. Besides, the model was proposed in three stages, supporting tailoring for applications beyond this thesis. The energy model was then applied to a theory of information processing efficiency by examining and extending existing theories. In particular, a new metric was introduced to express signal processing efficiency: the Landauer-Decibel. The first advantage of Landauer-Decibel is the reference to a physical limit. The second advantage is the logarithmic representation, predestined to express technological progress over time and compare large efficiency quantities.
====Download====
https://open.fau.de/handle/openfau/33406
https://doi.org/10.25593/open-fau-1404
https://tnotes.de/doc/Dissertation_Andreas_Tobola.pdf
""PhD"" Thesis / Doktorarbeit
Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), Technische Fakultät
2024
Autor: Andreas Tobola
====Abstract====
Wearable technology has been rapidly changing the overall picture of medical diagnosis. Highly integrated hardware and smart algorithms are enabling long-desired healthcare applications. Nevertheless, engineers still face major challenges. One such bottleneck is the limited energy of sensors worn on the body. Despite highly integrated hardware, the energy limitation requires larger batteries and eventually energy harvesting in addition.
This thesis gives a comprehensive insight into development of portable sensor technology, especially for biomedical applications, with a particular focus on minimizing energy consumption. One of the central drivers of this work is the philosophy of taking a holistic view of the entire signal processing chain, from the sensor to the evaluation of biomedical information. For this purpose, all parts of a biomedical signal processing chain were examined. The examination included hardware and software components.
Specialized tools and methods were developed for this purpose. A new power profiling method was developed to evaluate the power consumption of microcontroller-based systems. The unique feature of this method is the automated profiling tool which analyzes the system’s power consumption for a better understanding of the underlying processes. A further contribution of this thesis is new sensor hardware and new software. This sensor system was particularly designed for low-power evaluation. The sensor system simplifies low-power debugging with modularity and the comfort of dedicated power measurement terminals. Different variants of the sensor system hardware and software were developed and evaluated. The hardware variants investigated in this thesis were various sensor circuits, power supply circuits, and microcontroller types. Examined software variants were real-time operating systems, software-based power managers, and signal processing algorithms. The outstanding feature of the system is its scalability, achieved by designing software and hardware components configurable. The configuration determines the behavior of the sensor system. Also, the configuration influences power consumption. For the Ultra-low-power Sensor Evaluation Kit (ULPSEK), the average power consumption was determined between 50 µW and 10 mW, depending on the configuration.
The measurement results from the test bench, combined with theoretical principles, were merged into a general mathematical energy model to support the engineer in planning a new sensor system. Compared to state of the art, the model combines two standard sleep modes. Besides, the model was proposed in three stages, supporting tailoring for applications beyond this thesis. The energy model was then applied to a theory of information processing efficiency by examining and extending existing theories. In particular, a new metric was introduced to express signal processing efficiency: the Landauer-Decibel. The first advantage of Landauer-Decibel is the reference to a physical limit. The second advantage is the logarithmic representation, predestined to express technological progress over time and compare large efficiency quantities.
====Download====
https://open.fau.de/handle/openfau/33406
https://doi.org/10.25593/open-fau-1404
https://tnotes.de/doc/Dissertation_Andreas_Tobola.pdf
