Precision Tuning of an Accelerometer-Based Pedometer Algorithm for IoT Devices
Dorra Ben Khalifa, Matthieu Martel · 2021
Internet of Things (IoT) is considered as one of the most evolving technologies. It impacts almost every aspect of life in the modern world. One thing that prevents this technology from reaching its full potential is the too high memory and energy consumption of IoT devices. An aspect of this problem is the numerical computations performed by IoT devices which are generally carried out in higher precision than needed, by lack of techniques to tune this precision in function of the actual needs of the application. To surpass this limitation, we present our tool POP based on a static program analysis, done by abstract interpretation. POP provides a mixed precision tuning that finds the instructions and variables that may use lower precision with respect to the user accuracy requirements on the results. We demonstrate the efficiency of POP to tune an accelerometer-based pedometer algorithm for embedded applications.