Optimizing energy efficiency in distributed mHealth networks

Abstract

Energy efficiency in mobile health applications is a relevant problem for long-term monitoring and user acceptance. Various parameters influence the runtime of the system to some degree. One of the parameters is the sampling rate of the individual distributed sensors. Increasing the sampling rate can lead to an increase in energy consumption within the system. By contrast, a reduction can lead to a loss of the data quality, which reduces the informative value of the results of algorithms that use this data. Using optimization methods from reinforcement learning and deep learning to adaptive adjust the sampling rates during runtime, energy efficiency could be improved in only 40 training runs without losing data quality during sampling.