Algorithm 2 Classifier

Input: real-time data A

Output: human activity category Preprocess the input data through the Butterworth low-pass filter and dividing by sensitivity coefficient $k$. Segment the data into overlapping windows. Extract the features and construct the feature vector x with normalization. Input x into three MGD models and calculate $p_1\left(x\right),p_2\left(x\right)$, and $p_3\left(x\right)$. if $p_1\left(x\right)==\underset{i=1,2,3}{\max }{p}_i\left(x\right)$: Output (LIA) else if $p_2\left(x\right)==\underset{i=1,2,3}{\max }{p}_i\left(x\right)$: Output (MIA) else if $p_3\left(x\right)==\underset{i=1,2,3}{\max }{p}_3\left(x\right)$ & $p_3\left(x\right)>{\epsilon }_2$ & $TA<{\epsilon }_3$: Output (VIA) else:Output (Fall)