Answers are grounded in this paper and cite their sources.
Demonstrates improved modulation recognition accuracy in complex communication systems, suggesting enhanced algorithm efficiency.
With the advancement of Big Data technology, deep learning automatic modulation recognition (DLAMR) has undergone new improvements. Existing DLAMR methods focus mostly on the primary matching of the model itself or ubiquitous big communications data, which lack interpretability and ignore deep representations for the modulation mechanism of the communication signals; thus, difficulties in further improving the recognition accuracy and multiquadrant amplitude modulation (MQAM) discriminability in complex communication environments are encountered. In response to these challenges, this article proposes an innovative communication signal graph mapping method to address the uncertainty in the modulation mechanisms. Specifically, it models sampling points as nodes; connects inter- and intrasymbol points with edges to represent modulation mechanisms and propagation uncertainty; and maps amplitude, phase, in-phase, and quadrature values as node features. A deep graphical and temporal neuro-fuzzy methodology (GT-DNFS) that integrates graph attention networks and bidirectional long short-term memory networks is subsequently proposed for DLAMR. The numerical results show that GT-DNFS achieves a significantly higher recognition accuracy of 93.01%, and an MQAM (M=16, 64) discrimination of 94.5%. This research offers valuable insights for neuro-fuzzy networks and efficient DLAMR algorithm design.
No takes yet. Share an insight, caveat, or question.
Jian et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: