Thesis
Design of a low-power information-aware sampling architecture using learning-based feedback techniques
Washington State University
Master of Science (MS), Washington State University
05/2019
DOI:
https://doi.org/10.7273/000004223
Handle:
https://hdl.handle.net/2376/125119
Abstract
The exponential growth in the current digital economy is leading to a dramatic increase in data volumes. This is leading to large overheads in data transmission necessitating development of new sampling methods for transmitting wireless information from sensor nodes. This work proposes an information-aware adaptive sampling system architecture that can operate with ultra-low-power consumption. Mixed-signal circuits and system design paradigms are proposed to overcome the fundamental trade-off between power consumption and quality of decision-making using feature-based metrics. A sequency-based feature extraction with 1-bit binary antipodal matrix multiplication and a two-tier classification technique is introduced for adaptive sampling. The proposed scheme augments operation of existing analog-to-information converters, and can save up to 70% power through both data compression and transmission of relevant information. We compare the effects of finite precision of feature coefficients in the proposed architecture and demonstrate less than 2% classification error rate with only 8-bit precision in MIT-BIH database comprising of 20 patients.
Metrics
1 File views/ downloads
22 Record Views
Details
- Title
- Design of a low-power information-aware sampling architecture using learning-based feedback techniques
- Creators
- Arya Alex Rahimi
- Contributors
- Subhanshu Gupta (Advisor) - Washington State University, Electrical Engineering and Computer Science, School of
- Awarding Institution
- Washington State University
- Academic Unit
- Electrical Engineering and Computer Science, School of
- Theses and Dissertations
- Master of Science (MS), Washington State University
- Publisher
- Washington State University
- Identifiers
- 99900896431601842
- Language
- English
- Resource Type
- Thesis