Lightweight electrocardiogram signal compression

DOI

https://doi.org/10.1016/j.bspc.2023.105012
Forlagets udgivne version

Cardiovascular diseases are the primary cause of death around the world. Cardiovascular ailments can be monitored continuously using wearable devices, which are resource-constrained and battery-operated. Efficient data compression is a promising method to reduce transmission energy cost and extend battery lifetime. In this paper, we first propose two novel pruning methods with pruning matrix's entries taken from the set {−1, 0, 1}, which achieve similar reconstruction quality as compared to DCT pruning but with far less computational cost due to multiplierless operation. Reconstruction for these pruning methods is proposed using Gaussian elimination. Then we design two lightweight compression algorithms based on the pruning methods, i.e., sign compression algorithm and binary compression algorithm. The proposed compression algorithms and pruning methods are evaluated using performance metrics such as compression ratio, percentage root mean square difference, quality score, etc. Furthermore, The pruning methods are implemented in TelosB mote and execution time is evaluated. The experiments are conducted on the MIT-BIH arrhythmia database. The experimental results show that the proposed compression algorithms outperform the state-of-the-art methods.

Originalsprog	Engelsk
Artikelnummer	105012
Tidsskrift	Biomedical Signal Processing and Control
Vol/bind	85
Antal sider	11
ISSN	1746-8094
DOI	https://doi.org/10.1016/j.bspc.2023.105012
Status	Udgivet - aug. 2023

Find uddannelse

Studievejledning

Besøg AU

Kvalitet

Find

Aktuel forskning

Talent

Virksomheder

Kommuner og regioner

Gymnasier

Myndigheder

Organisation

Kontakt

Profil

Job

Om AU

DOI