A hybrid SOM and HMM classifier in a Fog Computing gateway for Ambient Assisted Living Environment

Abstract

Implementing Machine Learning algorithms with low-power devices and limited computational resources is challenging. Research on temporal data of Ambient Assisted Living (AAL) environment sensors to handle many states of the deep learning models that are used to train unsupervised data is limited. Computational aspects of the data training and inferring the insights from the sensor data of the AAL environment are essential aspects of a fog computing framework. The AAL environment embodies a fog computing structure with limited computing capabilities. This paper studies how to train and infer meaning information from the AAL sensor data using a hybrid algorithm of Self Organizing Map (SOM) and Hidden Markov Model (HMM) on a resource constraint computing device such as Raspberry Pi was explored. The research investigations reveal that the execution of the hybrid method on the fog computing gateway could cluster the anomalous instances accurately.