A Hybrid Data-Model Driven Method for Parameter Estimation of Typical Combined Deception Jamming

Radar, as a crucial component in modern electronic warfare, is frequently subjected to various forms of deception jamming. Accurate estimation of jamming parameters is essential for effective electronic countermeasures. Intermittent Sampling Repeater Jamming (ISRJ) and Smeared Spectrum (SMSP) jamming are two typical coherent jamming techniques. When coexisting, they are prone to aliasing in the time-frequency domain. Under such conditions, traditional parameter estimation methods, often based on a single jamming model, struggle to achieve high accuracy. Furthermore, model-driven neural network approaches can be constrained by the size of training datasets and real-time processing demands. To address these challenges, this paper proposes a hybrid data-model driven method for estimating parameters of combined deception jamming. For parameters with low degrees of freedom (DoF), a rapid estimation framework is developed by integrating classical signal processing techniques with insights into jamming mechanisms. Conversely, for parameters with high DoF, a Convolutional Neural Network (CNN)-based estimation module is designed. This network is pre-trained to meet real-time processing demands and achieve high-accuracy estimation.