Spectral fusion-based breathing frequency estimation; experiment on activities of daily living

Table 3 Table of results

Activity protocol	\(\mid S_{xy}\mid ^2\)		\(\mid S_{xx} \cdot S_{yy}\mid\)		\(S_{xx}\)		\(S_{yy}\)
Activity protocol	\(\%D^{2\sigma }\)	\(\%E\)	\(\%D^{2\sigma }\)	\(\%E\)	\(\%D^{2\sigma }\)	\(\%E\)	\(\%D^{2\sigma }\)	\(\%E\)
Office work	13.09	7.82	8.65	5.61	16.48	11.56	14.08	12.21
Emotional stress	13.62	10.98	9.03	8.68	16.47	11.67	16.70	14.42
Floor sweeping	13.43	11.08	10.74	9.03	22.17	15.77	18.75	11.57
Tidying up	17.96	13.74	14.05	11.50	25.37	17.94	21.25	14.18
Table cleaning	16.91	13.01	9.93	9.05	22.21	15.82	18.51	12.33
Walking	11.31	11.31	8.51	8.74	17.35	15.99	23.00	16.95
Cycling	8.10	9.44	5.40	6.96	21.35	16.97	10.10	11.71
Running	15.59	19.17	12.58	16.05	28.20	24.06	25.16	25.30
Average	13.75	12.06	9.86	9.45	21.20	16.22	18.44	14.83

The performance of RRI spectral-based (\(S_{xx}\)) and MSV spectral-based (\(S_{yy}\)) BF estimation as well as fusion methods, including CPSD-based (\(\mid S_{xy}(t,f)\mid ^2\)) and spectral multiplication (\(\mid S_{xx} \cdot S_{yy}\mid\) ) BF estimation over the protocols. For each method, average \(\%D^{2\sigma }\) and \(\%E\) is reported and the lowest error in each protocol is written in italics

ISSN: 1475-925X