Efficient Online Compression for MEMS based BCG Wearable Sensors on ULP FPGA

Publikationstyp
Conference Paper
Publikationsdatum
2023
Sprache
English
Author
Kulau, Ulf 
Ahmed, Abdelrahman Noshy Abdelalim 
Institut
Smart Sensors E-EXK3 
TORE-URI
http://hdl.handle.net/11420/15344
Citation
10th IEEE International Symposium on Inertial Sensors and Systems (INERTIAL 2023)
Contribution to Conference
10th IEEE International Symposium on Inertial Sensors and Systems, INERTIAL 2023 
Publisher DOI
10.1109/INERTIAL56358.2023.10103939
Scopus ID
2-s2.0-85159067585
Compression of Ballistocardiography (BCG) data is of a great importance specially in the context of wearables and ultra-low power (ULP) applications, respectively. This paper presents an efficient and yet simple compression core for BCG data that can be integrated to MEMS sensor or on ULP FPGAs. The proposed compression technique is a modified delta encoding algorithm that can compress data efficiently ranging from lossless to lossy compression, while the design was derived from BCG specific requirements. The technique offers flexibility with respect to compression performance and signal distortion where compression ratio can be traded for lossless compression and vice verse. Evaluations of 4 BCG data sets show an average compression ratio of 3 with adequate PRDN. This compression core is further implemented in VHDL and it utilizes 234 LUTs of FPGA resources supporting online compression.
Schlagworte
BCG
Data compression
FPGA
MEMS
Ultra-low power
Wearable