Skin Conductance Response in Real Driving Settings: Comparison of Analysis Methods

Human skin conductance, modulated by sweat secretion, is known to be a possible indicator of both attention and memory capabilities, which, in their turn, may influence the behavior of a driving person. In particular, the number of peaks appearing in the skin conductance signal, joint with their amplitude, can be used as a quantitative index to evaluate the driver's status. This paper presents an analysis of the different results that three literature methods, namely Continuous Decomposition Analysis (CDA), Discrete Decomposition Analysis (DDA), and Trough-To-Peak (TTP), output when applied on the same signals acquired during real driving sessions, both short and long ones, performed in uncontrolled settings of either urban and extra-urban traffic scenarios. The comparison of the numerical results shows that the TTP method usually underestimates the number of peaks, with respect to CDA and DDA, while the results are more similar in terms of mean amplitude of the skin conductance peak. The analysis also suggests to keep
$0.01\ \mu\mathrm{S}$
as the proper minimum amplitude criterion for the TTP method, when the sensor used provides low amplitude skin conductance signals, to avoid increasing the number of missed peaks.