TY - GEN
T1 - FitBeat
T2 - 2017 IEEE International Conference on Smart Computing, SMARTCOMP 2017
AU - Tu, Linlin
AU - Huang, Jun
AU - Bi, Chongguang
AU - Xing, Guoliang
N1 - Publication details (e.g. title, author(s), publication statuses and dates) are captured on an “AS IS” and “AS AVAILABLE” basis at the time of record harvesting from the data source. Suggestions for further amendments or supplementary information can be sent to [email protected].
PY - 2017/6/12
Y1 - 2017/6/12
N2 - Tracking heart rate for fitness using wrist-type wearables is challenging, because of the significant noise caused by intensive wrist movements. In this paper, we present FitBeat- A lightweight system that enables accurate heart rate tracking on wrist-type wearables during intensive exercises. Unlike existing approaches that rely on computation-intensive signal processing, FitBeat integrates and augments standard filter and spectral analysis tool, which achieves comparable accuracy while significantly reducing computational overhead. FitBeat integrates contact sensing, motion sensing and simple spectral analysis algorithms to suppress various error sources. We implement FitBeat on a COTS smartwatch, and evaluate the performance of FitBeat for typical workouts of different intensities, including walking, running and riding. Experimental results involving 10 subjects show that the average error of FitBeat is around 4 beats per minute, which improves heart rate accuracy of the default heart rate tracker of Moto 360 by 10x.
AB - Tracking heart rate for fitness using wrist-type wearables is challenging, because of the significant noise caused by intensive wrist movements. In this paper, we present FitBeat- A lightweight system that enables accurate heart rate tracking on wrist-type wearables during intensive exercises. Unlike existing approaches that rely on computation-intensive signal processing, FitBeat integrates and augments standard filter and spectral analysis tool, which achieves comparable accuracy while significantly reducing computational overhead. FitBeat integrates contact sensing, motion sensing and simple spectral analysis algorithms to suppress various error sources. We implement FitBeat on a COTS smartwatch, and evaluate the performance of FitBeat for typical workouts of different intensities, including walking, running and riding. Experimental results involving 10 subjects show that the average error of FitBeat is around 4 beats per minute, which improves heart rate accuracy of the default heart rate tracker of Moto 360 by 10x.
UR - http://www.scopus.com/inward/record.url?scp=85022335888&partnerID=8YFLogxK
UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-85022335888&origin=recordpage
U2 - 10.1109/SMARTCOMP.2017.7947009
DO - 10.1109/SMARTCOMP.2017.7947009
M3 - RGC 32 - Refereed conference paper (with host publication)
SN - 9781509065172
T3 - 2017 IEEE International Conference on Smart Computing, SMARTCOMP 2017
BT - 2017 IEEE International Conference on Smart Computing, SMARTCOMP 2017
PB - IEEE
Y2 - 29 May 2017 through 31 May 2017
ER -