The paper describes the comparison of activity recognition system consisting of a single accelerometer sensor and multiple accelerometer sensors. The paper also presented the influence of sampling rate towards recognition accuracies, most widely used features in this research area and the time take to calculate them, and comparison between classifiers in activity recognition. The result of the studies show that, sampling rate beyond 20Hz increases recognition accuracy by just 1%, also stabilized beyond 50Hz. Therefore, 20Hz frequency is proposed as the sampling rate for wearable system using multiple sensors. The mean, zero crossing rate, mean crossing rate and root mean square consumes are the least time take to compute which is 0.02ms. The standard deviation and variance took 0.05ms to compute. Tilt angle and angle velocity took more than 0.1ms to compute. Considering the recognition accuracy obtained from previous studies, it is concluded that mean and variance are proposed for wearable system using multiple sensors. Decision tree achieved second highest recognition accuracy in the study behind ANN. Considering the long training time is required for ANN classifier, Decision tree classifier is identified as an efficient classifier for wearable system. The result also further established the case for multi-sensor systems in activity recognition.
 L. Gao, A. K. Bourke, and J. Nelson, “Evaluation of accelerometer based multi-sensor versus single-sensor activity recognition systems,” Medical Engineering & Physics, vol. 36, no. 6, pp. 779–785, Jun. 2014.