A Machine Learning Based Energy-Efficient Indoor Multiple IoT Device Tracking Algorithm Based on Correlated Group Determination

Abstract

We develop a novel algorithm for energy-efficient indoor multiple IoT device tracking based on Artificial Intelligence (AI). Our algorithm is comprised of two phases: First we jointly forecast the future positions of the mobile IoT devices. Second we develop a novel algorithm that determines groups of IoT devices whose forecast trajectories are correlated with each other over a future time window. Our simulations demonstrate that our algorithm results in significant energy savings by keeping only the leader of the correlated group active while putting the followers to sleep during the entire duration for which the correlated group persists. This results in low intra-communication energy costs for the correlated group. This work represents a significant advance over single-device tracking algorithms by exploiting the correlations between the trajectories of multiple IoT devices.