Table 1 An overview of MA removal techniques

Pengfei et al. [25]

–

2 PPG, 3 ACC were sampled at 100 Hz

FT RLS filtering

Reflective type PPG sensor on the left hand reference

PPG was available via the right hand

Fallet and Vesin [26, 27]

Downsampled all signals to 35 Hz

2 PPG, 3 ACC were sampled at 125 Hz

NLMS

IEEE SPC, 2015

All six combinations of PPG and ACC signals were used

Wood and Harry [30]

–

1 PPG, 3 ACC signals were sampled at 100 Hz

Laguerre-based ANC

Ring type PPG sensor

Jogger experiment

Ram et al. [37]

–

1 IR, 1 red PPG signals were sampled at 250 Hz

Adaptive step LMS filtering

Synthetic MA generation using FFT, SVD, and ICA

Tested for randomness using skew and Kurtosis

Yalan Ye. [42]

Signals were bandpassed using 0.4–5 Hz

2 PPG, 3 ACC were sampled at 125 Hz

Non-linear Volterra-based RLS adaptive algorithm

TROIKA [65], IEEE SPC, 2015

–

Yangsong Zhang [51]

Signals were bandpassed using 0.4–5 Hz

2 PPG, 3 ACC were sampled at 125 Hz

Correlation-based MA removal using EEMD

Fast EEMD (FEEMD), IEEE SPC, 2015

–

Biagetti et al. [58]

Signals were bandpassed using 0.4–5 Hz

2 PPG, 3 ACC were sampled at 125 Hz

SVD-based decomposition of signal present in Hankel space

IEEE SPC, 2015

Exercise model

Kim and Yoo [59]

–

1 IR, 1 red PPG signals were sampled at 500 Hz

ICA decompostion

–

–

Ziling Zhang [66]

Downsampled signals to 25 Hz and signals were bandpassed using 0.4–4 Hz

2 PPG, 3 ACC signals were sampled at 125 Hz

JSSR using MMV model

IEEE SPC, 2015

Rojano and Isaza [47]

–

2 PPG, 3 ACC signals were sampled at 125 Hz

SVDTFD and DWT and EEMD

11 signals taken IEEE SPC, 2015

–

Dubey et al. [69]

–

2 PPG, 3 ACC signals were sampled at 125 Hz

HSUM

IEEE SPC, 2015

–

Schack et al. [70]

Downsampled signals to 25 Hz and signals were bandpassed using 0.5–6 Hz

2 PPG, 3 ACC signals were sampled at 125 Hz

Adding squared PPG spectra and Gaussian filtering

IEEE SPC, 2015 indicator function which showed correlation-based fundamental frequency