Trajectory Tracking Approach Based on Mode Decomposition Under Malicious Attacks

The accuracy of tracking methods that are widely used today is often impressive. These methods are firmly grounded in the assumptions of Markov models and require meticulous prior model information. However, when they are applied to scenarios that involve non-cooperative and highly maneuvering targets, these prerequisites frequently become unattainable. To overcome this constraint, a trajectory fitting model, which incorporates both trend and periodic terms to improve the precision of target trajectory representation, is introduced based on the fundamental principle of mode decomposition. It enhances the adaptability and robustness of the tracking system. The proposed method greatly reduces the dependence on prior information and provides an efficient and interpretable technical approach for tracking maneuvering targets. To enhance data security against eavesdropping attack, a noise-injection encryption method is proposed by integrate scalarization. The information density is lowered through scalarization, while the injected noise obscures the original features and maintains consistent data dimensions. This combination significantly enhances the confidentiality of data transmission. Furthermore, since the Gaussian assumption is overly stringent and often unfulfilled in practical scenarios, a detection scheme is proposed, which integrates the Jensen-Shannon (JS) divergence with the K-Nearest Neighbor (KNN) algorithm. This scheme improves precision and efficiency by computing minimum Euclidean distances with KNN and leveraging JS divergence. Simulation experiments demonstrate the efficacy of the proposed algorithm.