Abstract
Human activity recognition (HAR), driven by machine learning techniques, offer the detection of diverse activities such as walking, running, and more. Considering the dynamic nature, limited energy and mobility of wireless body area networks (WBANs), HAR can play a significant role in enhancing WBANs performance. This paper genuinely bridges HAR's activity recognition capability using machine learning to develop a novel WBAN routing decisions adoptively. Being optimum in power consumption, we employed Random Forest classification algorithm for activity recognition. The resulted system holds great promise for optimizing routing decisions, improving energy efficiency, and enhancing the overall performance of WBANs in healthcare and related domains. To evaluate the performance of the proposed protocol, we have measured various performance metrics, including energy consumption, throughput, and the number of dead nodes. The results have been compared with mobTHE protocol to demonstrate the effectiveness of our HAR based Routing protocol.