End-to-end radar signal sorting based on semantic segmentation
Tao Chen, Jiashuai Li, Limin Guo, Lei Yu · IET conference proceedings. · 2024
This research offers a semantic segmentation network-based signal clustering method to overcome the drawbacks of traditional radar signal sorting algorithms, which mainly rely on manually setup parameters and are limited by algorithm structures. Through this approach, end-to-end radar signal clustering is made possible by removing the aforementioned limits. This process uses the PDW of the combined radar pulses to produce a dot matrix image. DeepLab V3+ semantic segmentation network can be used to achieve pixel clustering. Then map the clustered pixels back to the pulse stream. It is evident from the simulation results that this method works well at clustering signals and maintains strong performance even in environments with pulse loss. Also, our method shows better accuracy when compared to conventional sorting algorithms.