A Semi-Supervised Deception Jamming Discrimination Method

For anti-deception jamming discrimination on multistatic radar system, the existing anti-jamming methods based on artificial intelligence require enough training samples with a large amount of labeled data, but it is difficult to obtain lots of labeled radar echo data in reality combat environment. This paper proposes a convolutional deep belief network-based deception jamming discrimination method for the insufficient labeled data. The constructed anti-jamming network is trained with a large number of unlabeled radar echo data, and enhances the discriminating capability of the network with a small number of labeled echo data. It realizes a more accurate deception jamming discrimination network, which achieves full information utilization and broadens the limitation condition of jamming discrimination. Simulation results show that compared with the existing artificial intelligence-based jamming discrimination method utilizing tens of thousands of labeled data, the proposed method satisfies the same performance utilizing 2000 labeled data. It reduces data requirements and enhances operational capabilities.