Wearable IoT power consumption optimization algorithm

IoT devices used in eHealth applications are rapidly advancing towards being more compact, performant and low-powered. One of the contemporary challenges in technology is represented by the limitations of batteries in terms of weight and energy capacity. Particularly for eHealth devices, this challenge commonly becomes one of the main characteristics of the device. As most eHealth devices have been adapted to use BLE (Bluetooth Low-Energy) communication to achieve the high energy autonomy, there is an increasing trend for wearable devices that do not rely on a smartphone and can independently process and send data through LPWAN (Low Power Wide Area Network) type networks like NB-IoT and LTE-M (Long Term Evolution category M). In this paper we present an algorithm which leverages Machine Learning model predictions, to achieve lower power consumption when using LPWA modems for IoT devices. While the algorithm can be universally applied to any device, we focused on eHealth devices for which the cellular signal greatly varies in time (situations where a patient travels in and out of urban areas).