Algorithm 1. Raw acceleration filtering and transformations, and STS possible location identification. RM is rolling mean, CWT is continuous wavelet transform.

Data: $y_{acc}=[y_{ax},y_{ay},y_{az}]$ Result: a, $a_{filtered}$, $a_{RM}$, P, $ST{S}_{possible}$ $a\leftarrow \sqrt{y_{ax}^2+y_{ay}^2+y_{az}^2}$; $a_{filtered}\leftarrow butterworth\left(\right|a|,order=4,cutoff=5\mathrm{Hz})$; $a_{RM}\leftarrow RM(a_{filtered},window=0.25\mathrm{s})$; $c\left(F\right),F\leftarrow CWT(a_{RM},gaus1,scale=1\to 64)$; $P\leftarrow {\sum }_{f=0\mathrm{Hz}}^{0.5\mathrm{Hz}}{c}_i\left(f\right)$; $ST{S}_{possible}\leftarrow peaks(P,height={\sigma }_P,distance=1\mathrm{s})$;