Data Efficient PV based Indoor Event Detection

In this paper, we argue that instead of indiscriminately transmitting every sensed value, data transmission should prioritize relevance, intelligence, and efficiency. As a case study, we investigate exit/entry event detection through the generation of signature voltage patterns from PV module connected to an IoT platform and placed near entrance. This detection technique facilitates automated appliance management (e.g., lighting and HVAC control) and improves indoor security (e.g., unauthorized entry detection) by leveraging existing devices for sensing purposes. To enhance data efficiency, we propose the implementation of an intelligent on-board event detection algorithm, prioritizing sampling points within specific timeframes based on their significance on classification accuracy and lossless compression technique to further squeeze necessary information for classification. In our testbed evaluation, our approach successfully achieved a 99.7% data compression rate, substantially reducing the number of required advertisements and extending operation within a fixed energy budget.