Abstract
The Wireless sensor networks (WSNs) monitor the dynamic environments that change rapidly over
time. In this paper Real-time focus for activity recognition using WSN that accurately detects gestures and
body movements. Initially, the nodes, which are distributed throughout the body, detect the movement of the
organ through an accelerometer sensor with measurements of three axes (positive, negative and null), where
these measurements are used by a hidden Markov model (HMM) to predict activity on each sensor.
Sensor activation and selection are based on the sensor's potential contributions to the classifier's accuracy
(ie, select the sensors that provide the most informative description of the gesture). To generate a final gesture
decision, a naive Bayesian classifier is used to combine the predictions of independent nodes to maximize the
posterior probability of Bayes' theorem.
