Long short‐term memory‐based power‐aware algorithm for prompt heterogenous activity

Abstract

SummaryHuman–computer interaction plays a vital role in wireless body area networks, internet of things, and big data. Wearables are low‐power devices with minimal battery capacity. In general, wearables suffer from energy losses due to changes in the user's body posture, diffraction, reflection, and shadowing of the human body. As a result, many control packets are needed to ensure proper communication in wireless body area network. Hence, this research proposes a long short‐term memory‐based power‐aware (LSTM‐PA) algorithm to ensure burst data transmission during prompt heterogeneous activities in the presence and absence of inter‐WBAN interference. The proposed algorithm predicts the best quality time (BQT) for data transmission by activity classification and robust R2 similarity (RRS) metric. The minimum transmission power is estimated by the critical point classification technique. The activity classification accuracy is 92% in the LSTM‐PA algorithm. The energy consumption in the node is reduced by up to 46.34% compared to benchmark algorithms.