Recognition， Inference and Prediction of Advanced Multi-Function Radar System Behaviors： Overview and Prospects

‍ ‍Multi-function radars （MFR） are typical representatives of advanced radar emitters. MFRs are capable of scheduling multiple simultaneous tasks over multiple targets along the radar timeline. The characteristics of the MFRs include instantaneous beam direction scheduling， flexible and dynamic work modes and complex signal modulation. With the continuous development of computational intelligence， cognitive theory， digital arrays， software-defined and hardware-reconfigurable systems， the degree of freedom and performance potential of advanced multi-function radar systems continue to increase， which poses great challenges to modern electronic reconnaissance and countermeasures systems. This paper focused on the perception and recognition challenges of the advanced MFR system behaviors and conducted a comprehensive survey from followings four aspects： i） modeling and characterization of the reconnaissance analysis model for analyzing and identifying the behavior of multifunctional radars； ii） recognition of radar system’s explicit behaviors； iii） inference of radar system’s implicit behaviors， and iv） prediction of future system behaviors. Firstly， this article provided an overall categorization of radar system behaviors. Based on the content and purpose of radar system’s behavior activities， radar behaviors were divided into three categories： inference behavior， decision behavior and action behavior. The above three types of behavior were then classified and divided from two different perspectives： from the radar’s side with management behavior and assessment behavior； from the reconnaissance system’s side with explicit behavior and implicit behavior. The relationship between the two perspectives and three categories were analyzed. Secondly， the reconnaissance analysis model was proposed for the first time. The multi-function radar reconnaissance analysis model is a radar behavior model constructed with the multi-function radar ontology as the object. This model starts directly from the behavior of the multi-function radar ontology， taking the observable explicit behavior from the reconnaissance system’s side as the inputs， integrating all available information that affects the radar behavior， and aims to recognize the attributes， generation mechanism， action process， behavior results of the multi-function radar behavior， and also acquiring subsequent behavior utilization and countermeasure methods. Thirdly， a comprehensive survey regarding the recognition of radar system’s explicit behaviors， inference of radar system’s implicit behaviors， and prediction of future system behaviors was conducted. The problem formulation and research route of each task were first analyzed and then corresponding researches were reviewed according to the proposed research route. Finally， this paper analyzed the challenges， priority of future research and drawn conclusion to provide useful information for subsequent theoretical research and technological development related to advanced MFR system behavior and concluded the whole article. These future directions were concluded from five aspects including i） behavior modeling of advanced MFR systems； ii） general large model method in electromagnetic domain； iii） fast and accurate inverse signal processing and reward function estimation methods； iv） efficient countermeasures against advanced MFRs based on the behavioral game theory； v） theoretical error analysis of electronic reconnaissance tasks and methods； vi） general behavioral benchmarks datasets for performance comparison of different methods.