%0 Journal Article
%A DUAN Rui-Feng
%A LI Cheng-Ju
%A YANG Jia-Chen
%T <span>CNN demodulation model with cascade parallel crossing&nbsp;</span><span>for CPM signals</span>
%D 2024
%R 10.19682/j.cnki.1005-8885.2024.1005
%J Journal of China Universities of Posts and Telecommunications
%P 30-42
%V 31
%N 3
%X <span>The continuous phase modulation (CPM) technique is widely used in range telemetry due to its high spectral</span><br>
<span>efficiency and power efficiency. However, the demodulation performance of the traditional maximum likelihood</span><br>
<span>sequence detection (MLSD) algorithm significantly deteriorates in non-ideal synchronization or fading channels. To</span><br>
<span>address this issue, this work proposes a convolutional neural network (CNN) called the cascade parallel crossing</span><br>
<span>network (CPCNet) to enhance the robustness of CPM signals demodulation. The CPCNet model employs a multiple</span><br>
<span>parallel structure and feature fusion to extract richer features from CPM signals. This approach constructs feature</span><br>
<span>maps at different levels, resulting in a more comprehensive training of the model and improved demodulation</span><br>
<span>performance. Simulation results show that under Gaussian channel, the proposed CPCNet achieves the same bit</span><br>
<span>error rate (BER) performance as MLSD method when there is no timing error, but with 1/4 symbol period timing</span><br>
<span>error, the proposed method has 2 dB demodulation gain compared with CNN and convolutional long short-term</span><br>
<span>memory deep neural network (CLDNN). In addition, under Rayleigh channel, the BER of the proposed method is</span><br>
<span>reduced by 5% -87% compared to that of MLSD in the wide signal-to-noise ratio (SNR) region.</span>
%U https://jcupt.bupt.edu.cn/EN/10.19682/j.cnki.1005-8885.2024.1005