KNN-Based Decision Model for Device Argumentation in Ambient Assisted Living

Abstract

AbstractAging population has increased in the past few decades with the rise in social and economic challenges. In particular, people wish to live an independent life without giving most of the burden to caretakers. This is where the Ambient Assisted Living (AAL) has gained momentum that allows elderly people to lead an independent and safe lifestyle by performing their daily activities. Such an AAL environment which consists of heterogeneous sensors and actuators helps in monitoring the daily activities carried out by elderly people throughout the day. These smart objects interact among themselves by using higher-level information generated as a result of processing the sensor values on chips to decide their own course of actions toward their goals. Thus, these smart objects which are now speaking and hearing objects (“things”) interact through argumentation. These objects could communicate and argue with each other about a particular activity in progress and to find the understanding of the present state and take decisions and act accordingly. Thus argumentation as a common basis for interactions among these smart objects, this paper focuses on implementing a KNN-based decision model for device argumentation about an activity identification and to take actions accordingly. The proposed solution shows that KNN gives an accuracy of 72.28%, Precision of 85%, Recall of 42.3% and F1-score of 38.40% for classifying ambiguous device argumentation in AAL.KeywordsAmbient Assisted LivingDevice argumentationDecision modelMachine learningActivity