An Improved Kalman Filter Measurement Model Employing Regularized Least-Squares Problem

Abstract

Unmanned Aerial Vehicles’ (UAVs) technology has recently had a major impact on our daily lives. Drone detectors are more important. This is not only due to the availability of service infrastructure but also due to the protection from serious security vulnerabilities. Drones are inherently small, fast, and can operate in low-altitude environments, allowing mobile or fixed radars to track UAVs. Methods based on the Kalman filter (KF) are used to extract valuable trajectory information from samples that contain large amounts of noise information. Because the drone’s trajectory resembles an uncertain behavior, traditional radiofrequency-based methods have lower tracking accuracy. Existing work namely Diffusion Map (DMK) was recently introduced to model uncertainty in an environment without prior knowledge. If the diffusion map working in a noisy environment, the model will be less accurate. To address this, this work present Uncertainty and Error Aware Kalman Filter (UEAKF) to track drones and bring the tradeoffs between past estimates and future measurements in a dynamic environment. Experimental results show that the UEAKF-based UAV monitoring model achieves better Root Mean-Square-Error (RMSE) performance than the traditional particle filter and the DMK-based UAV monitoring model.