An Extreme Edge Low Power Device with Wavelet-based Compression for Physiological signals

The combination between artificial intelligence, affective computing, and the internet of things, their use in real-time continuous monitoring applications, and their potential relevance in topics such as e-health or marketing has spotlighted the need of collecting extensive amounts of data constantly. However, at the same time, these wireless body sensor networks should assure system usability and user comfort in terms of battery lifetime. This is where data compression techniques appear. This paper presents a detailed design and embedded implementation of a digital wavelet transform (DWT)-based filter bank within an extreme edge low-power device. Besides, a study on the improvement in power consumption, considering memory capabilities, computational power resources, and data transmission rate, has provided a reduction of up to 60% in the energy required for wireless transmission.