Jamming Suppression of Corner Reflector via a Statistically Attentive TCN-SRU Network on Complex-Valued HRRPs

The letter presents a jamming suppression algorithm designed for dilution jamming of corner reflectors against seeker radar, which mitigates jamming based on complex-valued high-resolution range profile (HRRP) sequences. The algorithm integrates a temporal convolutional network (TCN) and a simple recurrent unit (SRU) layer. The TCN first extracts global features across the range dimension, while the SRU subsequently models long-term dependencies in both range and slow-time dimensions. To make advantage of information in complex data, the complex-valued HRRPs are processed through a complex-valued convolutional layer at the front end of TCN. The intermediate representation between TCN and SRU is weighted by statistical attention, facilitating comprehensive feature utilization. Evaluated on measured data, the algorithm gains a significant improvement in signal-to-interference ratio (SIR).