Semantic Flow Control for Task-Oriented Position Tracking

Abstract

Scalable position tracking systems are essential for ensuring efficient operation in dynamic and resource-constrained environments, particularly in applications requiring real-time accuracy and adaptability. Ultra-Wide-Band (UWB) communications facilitate high tracking accuracy with high beacon rate. However, this limits the number of parallel position tracking applications due to reduced communication effectiveness. We solve this problem by introducing Semantic Feedback to the mobile tag. By incorporating prediction accuracy into the Semantic Feedback, the mobile tag can adjust its beacon interval, reducing the frequency of transmissions when the localization system has high prediction confidence. To further improve this confidence, mobile tags also send their IMU (Inertial Measurement Unit) data to the Position Tracking System (PTS). By controlling the information flow in a task-oriented manner without incorporating IMU data, Semantic Flow Control improves communication effectiveness by 19.3% with slight reduction in localization accuracy, compared to using only a Kalman Filter. When integrated with IMU data, communication effectiveness further increases by additional 16.2% while transferred data decreases by 19.4%.