Balancing power consumption and data analysis accuracy through adjusting sampling rates: Seeking for the optimal configuration of inertial sensors for power wheelchair users

Abstract

Smartphones have already been used to capture wheelchair maneuvering data to analyze a wheelchair user’s activity level, which is directly related to his/her quality of life. Typically, the inertial sensors (e.g., accelerometer and gyroscope) in a smartphone are used for data collection. However, the limited battery life of the smartphone has become a major barrier to effective data collection. The sampling rate, as a primary configurable parameter of an inertial sensor, may have important impact on power consumption. Presumably, a lower sampling rate would consume less battery power. However, it may compromise the accuracy of data analysis. In this study, we investigate how the sampling rate of inertial sensors impacts the battery power consumption as well as the accuracy of data analysis. The four pre-defined sampling rate settings of the Android OS were evaluated for their impact on the smartphone’s power consumption. Additionally, we also measured the accuracy differences of the four sampling settings by comparing the sensor data-derived wheelchair maneuvering distances with the actual distances. The experimental results showed that it is possible and practical to balance the power consumption and data analysis accuracy by switching between appropriate sampling rate settings.