An Ultra Low Power Granular Decision Making Using Cross Correlation: Minimizing Signal Segments for Template Matching

Wearable sensor platforms have proved effective in a large variety of new application domains including wellness and healthcare, and are perfect examples of cyber physical systems. A major obstacle in realization of these systems is the amount of energy required for sensing, processing and communication, which can jeopardize small battery size and wear ability of the entire system. In this paper, we propose an ultra low power granular decision making architecture, also called screening classifier, that can be viewed as a tiered wake up circuitry. This processing model operates based on simple template matching. Ideally, the template matching is performed with low sensitivity but at very low power. Initial template matching removes signals that are obviously not of interest from the signal processing chain keeping the rest of processing modules inactive. If the signal is likely to be of interest, the sensitivity and the power of the template matching blocks are gradually increased and eventually the microcontroller is activated. We pose and solve an optimization problem to realize our screening classifier and improve the accuracy of classification by dividing a full template into smaller bins, called mini-templates, and activating optimal number of bins during each classification decision. Our experimental results on real data show that the power consumption of the system can be reduced by more than 70% using this intelligent processing architecture. The power consumption of the proposed granular decision making module is six orders of magnitude smaller than state-of-the-art low power microcontrollers.