A Waveform Optimization for an X-Band GaN Dual-Function Radar and Communication Front-End With Notch Filters and NN-DPD

We present an X-band dual-function radar-communication (DFRC) front-end with jointly optimized waveform and hardware. Two-notch filters at the T/R-switch antenna port suppress the second- and third-order harmonics, reducing the peak-to-sidelobe ratio (PSLR) by 1.3 dB. Implemented in a 250-nm gallium nitride (GaN) high-electron mobility transistor (HEMT) process, the integrated transmitter delivers saturated power of 41.5 dBm and peak power added efficiency (PAE) of 31.8% both at 9.5 GHz. The receiver provides 35.2-dB gain at 9.5 GHz and a minimum noise figure (NF) of 2.04 dB at 8.5 GHz. To synthesize radar-friendly communication sequences, we propose a hierarchical block-partitioned optimizer (HBPO) with an enhanced fast Fourier transform (FFT) for 16-quadrature amplitude modulation (16-QAM). With a 216-MHz radar band and a 4.567-ms coherent processing interval (CPI), the waveform achieves 0.693-m range resolution and 3.46-m/s velocity resolution at 9.5 GHz. Concurrently, a 64-QAM link over 629 MHz attains about 3.53 Gb/s gross bit rate. Measurements show HBPO improves PSLR by 7.5 dB on average and reduces search time by up to 29.6%. To linearize the realized GaN front end, neural-network-based digital predistortion (NN-DPD) is employed, which reduces error vector magnitude (EVM) by 9.6 dB and improves adjacent channel power ratio (ACPR) by 15.1 dB, while the localized discrete Fourier transform-spread-orthogonal frequency division multiplexing (DFT-s-OFDM) improves peak-to-average power ratio (PAPR) by 4.0 dB without degrading radar metrics. The demonstrated co-optimization delivers high-resolution sensing and high-fidelity communications on a single compact radio frequency (RF) chain, applicable for 6G DFRC front-ends.