Radar Model Recognition Based on Cascaded Markov Chain Model Under Missing Data Conditions

Radar model recognition constitutes a crucial component of electronic intelligence reconnaissance. Modern multifunctional radars (MFR) exhibit wide operating bandwidths and employ complex and diverse transmitted waveform parameters. However, constrained by their performance limitations, reconnaissance equipment often fails to comprehensively acquire complete waveform parameters of radar emitters. Moreover, the similarity in waveform parameters and patterns among different radar models further complicates radar model recognition. To address these challenges, this paper proposes a radar model recognition method based on a cascaded Markov chain model. By establishing Markov state transition probability matrices to characterize radar waveform switching patterns, the method first determines waveform patterns based on waveform unit transition rules, then identifies radar models through analyzing pattern transition regularities, thereby resolving recognition difficulties caused by parameter and pattern similarities. The Dempster-Shafer (DS) evidence theory is introduced for waveform pattern determination, enabling joint inference through multiple transition probability matrices to mitigate incomplete parameter acquisition caused by missing waveform units. Simulations demonstrate that the proposed method effectively identifies radar waveform patterns even with 80% missing waveform units under similar parameter conditions. Furthermore, when radar models share identical waveform patterns with 25% pattern missing rate, it achieves 80% recognition accuracy by leveraging differences in pattern transition patterns.