Optimizing smartphone energy consumption in sensor data collection and data transmission

Abstract

Sensors that come built in smartphones are extensively used by various apps and for analysis purpose. Some of these sensors, especially GPS and Microphone, are energy expensive. Hence, there is a need to use these judiciously. To study the trade-off between data accuracy and energy consumption due to sensor usage, we experiment with duty cycling while collecting sensor data. In many applications, the sensor data is required to be sent to a cloud or a server for storage or further computations. The size of the data is correlated to the amount of energy consumed in data transmission as the number of bytes to be transferred contributes significantly to the amount of energy consumed in data transfer. The key here is to reduce the size of data but at a minimal cost of accuracy. We propose a representative clustering approach to represent the dataset and transmit these representatives in place of the entire dataset. Further, we evaluate the amount of energy conserved by using this approach by transmitting representative data over different wireless technologies.