Dynamic sample rate adaptation for long-term IoT sensing applications

In long-term sensing applications data patterns can vary significantly over time. Often a multitude of sensors are used to measure different types of environmental conditions. Considering such variations it is hard to select a predefined sample rate that guarantees both, high data quality and energy efficiency. Hence, this paper presents a dynamic sample rate adaptation that strikes a balance offering optimal energy efficiency while maintaining high data quality. Based on the general concept of Bollinger Bands, a metric is derived that solely depends on the trend of the measured data itself. A real world measurement in the area of smart farming is used to show the effectiveness of this approach.