Covariance Intersection for Robust Sensor Fusion in Anti-drone Defense System

Abstract

In this study, we aim to establish a radar sensor-based defense system against the threat of suicide drones, which are not limited to attacks on personnel but can also neutralize critical friendly assets, including armored vehicles. Since drone tracking relies heavily on sensor performance, sensor fusion is employed to prepare for situations in which sensor performance is degraded, such as jamming attacks or sensor failures. First, an Unscented Kalman Filter (UKF) is used to estimate the position of an enemy drone in X, Y, and Z coordinates based on range, elevation, and azimuth measurements from radar sensors. The position estimation performance is further improved through sensor fusion using the Covariance Intersection (CI) method. To address sensor degradation, an Adaptive Kalman Filter (AKF), which updates the R matrix of the UKF, and a Fault Detection and Isolation (FDI) algorithm are applied to suppress the influence of degraded sensors. Simulations are conducted under assumed sensor-degradation scenarios, such as Noise Modulation Jamming and Range Deception Jamming, affecting a specific sensor. The results demonstrate that the proposed method maintains excellent position-estimation performance even under such conditions and enables interception of enemy drones using the Proportional Navigation Guidance (PNG) algorithm.