| 摘要: |
| 针对星载接收机接收到的船舶自动识别系统(Automatic Identification System,AIS)信号信噪比低、高斯最小频移键控调制采用的高斯成形滤波器产生码间干扰(Inter-Symbol Interference,ISI)以及非相干解调鲁棒性差等问题，提出了一种基于2比特判决反馈的改进AIS非相干解调算法。首先，接收信号通过匹配滤波器、鉴频器和自适应检测滤波器获得基带信号；其次，构造判决反馈，补偿鉴频噪声与ISI；最后，设计一种基于长短时记忆(Long-short Term Memory,LSTM)与注意力机制(Attention)的相位路径预测模型，获取2比特反馈补偿，改进判决反馈补偿能力。仿真与实测结果表明，改进方法较1比特判决反馈法对自适应滤波器的适应能力更强，在信噪比为7.2 dB时，误码率降低至10-5；较传统方法，真实场景下AIS信号的循环冗余校验通过率提升了9.5%。 |
| 关键词: 船舶自动识别系统(AIS) 非相干解调 判决反馈 LSTM-Attention |
| DOI：10.20079/j.issn.1001-893x.230719003 |
|
| 基金项目:四川省重点研发计划项目(2020YFG0051);校企合作项目(21H1445) |
|
| An Improved AIS Incoherent Demodulation Algorithm Based on 2-bit Decision Feedback |
| YUAN Ruichang,GONG Xiaofeng,XIAN Guo |
| (1.College of Electrical Engineering,Sichuan University,Chengdu 610065,China;2.Chengdu Dagong Bochuang Information Technology Co.,Ltd.,Chengdu 610097,China) |
| Abstract: |
| For the problems of low signal-to-noise ratio(SNR) of ship automatic identification system(AIS) signals received by satellite receivers,poor robustness of inter-symbol interference(ISI) and incoherent demodulation caused by Gaussian filtered minimum shift keying(GMSK) modulation using Gaussian shaped filters,an improved AIS non-coherent demodulation algorithm based on 2-bit decision feedback is proposed.Firstly,the received signal obtains baseband signal through matched filter,frequency discriminator and adaptive detection filter.Secondly,a decision feedback is constructed to compensate discrimination noise and ISI.Finally,a phase path prediction model based on LSTM and Attention is designed to obtain 2-bit feedback compensation and improve the decision feedback compensation capability.The simulation and experimental results show that the improved method has stronger adaptability to adaptive filters compared with the 1-bit decision feedback method.When the SNR is 7.2 dB,the bit error rate(BER) of the improved method is reduced to 10-5.Compared with that of traditional methods,the pass rate of AIS signal cyclic redundancy check（CRC） in real scenarios is increased by 9.5%. |
| Key words: automatic identification system(AIS) non-coherent demodulation judgment feedback LSTM-Attention |
