Electrodermal Activity Smart Sensor Integration in a Wearable Affective Computing System

Nowadays, the integration of smart sensors in wearable affective computing systems aims to improve and create applications based on conventional sensing technology. The inclusion of this type of sensors is key in high sensing-demanding applications, such as those based on affective computing. This sensing integration must be quantitatively supported by metrics that directly affect performance, such as processing time, memory usage, and measurement accuracy. This paper presents a comprehensive analysis of a specific smart sensor integration in a wearable constrained device from an embedded perspective. The authors performed this implementation by considering BINDI, which is a wearable device developed by the UC3M4Safety research group to prevent gender violence situations. BINDI monitors user physiological and physical variables to detect fear through artificial intelligence algorithms. Along with other sensors, this device incorporates an electroDermal activity (EDA) sensor. This type of sensor was proven to be a valuable source of information directly related to the autonomous nervous system (ANS) activation, which regulates physiological responses in stress and relaxed affective states. In the initial version of Binode, which was validated in different experiments with different volunteers, the EDA sensor is based on a DC measurement schema. However, AC measurement EDA circuitry is known to provide a higher amount of information. Thus, this work presents a detailed comparison between the current EDA sensor and a specific AC smart EDA sensor for a future version of BINDI.