A Radar Interference Signal Type Identification Method Based on Improved SqueezeNet
Yandi Luo, Ming Fang · 2024
Targeting the issues of large model size and high training costs in current radar interference recognition, this paper presents a method for recognizing radar interference signal types based on an enhanced version of SqueezeNet. A smooth pseudo-Wigner-Ville distribution is used to extract the time-frequency features of interference signals, and an improved lightweight SqueezeNet network model is used for intelligent recognition. In this paper, the mathematical models of 12 typical interference signals and their time-frequency feature extraction methods are described in detail, and the improved network structure characteristics of SqueezeNet are introduced. The experimental results show that the proposed method can significantly reduce the number of model parameters and training time on the premise of ensuring the recognition accuracy. Therefore, this method not only can effectively identify complex jamming signals, but also is especially suitable for resource-limited hardware platforms, which provides a new solution for anti-jamming technology of radar systems.