Table 1.
Comparison of existing schemes.
Fig 1.
Deep learning-based AFDM system architecture.
Fig 2.
Probability density function of the AFDM signal amplitude.
Fig 3.
Impact of the conventional companding transform method on the signal.
Fig 4.
PolyNet-Volterra Network Architecture.
Table 2.
Simulation parameter configuration.
Fig 5.
Convergence process during training under a fixed SNR (20 dB) environment.
Fig 6.
Convergence process during training under a dynamic SNR (5 dB to 25 dB) environment.
Table 3.
Benchmark model configurations and comparison rationale.
Table 4.
Computational complexity and performance scores of different models ().
Table 5.
BER of each scheme under different SNRs (0-25 dB).
Fig 7.
Bit Error Rate (BER) performance comparison under different SNRs (0-25 dB).
Table 6.
Feature-subset models.
Fig 8.
Comparison of reconstruction MSE in the masked region across different feature combinations.
Fig 9.
Histogram of parameter efficiency at 15 dB.
Fig 10.
Heat map analysis of the correlation between various input features and nonlinear distortion errors at a 10 dB signal-to-noise ratio.
Fig 11.
Heat map analysis of the correlation between various input features and nonlinear distortion errors at a 15 dB signal-to-noise ratio.
Fig 12.
Heat map analysis of the correlation between various input features and nonlinear distortion errors at a 20 dB signal-to-noise ratio.
Fig 13.
Scatter plot of the reconstruction error in the masked region versus the original signal power (SNR = 15 dB).
Fig 14.
Gradient magnitude and relative contribution of the input features.
Fig 15.
Feature weight distribution of the first-layer hidden neurons.
Fig 16.
Mean square error of various schemes at the masked positions.
Table 7.
Repeated independent Monte Carlo variability at 20 dB.
Table 8.
Uncertainty quantification with BER and masked-region MSE confidence intervals.
Table 9.
Formal significance tests against PolyNet-Volterra at 20 dB.
Fig 17.
Extended statistical validation at 20 dB, including repeated-run BER distributions, Wilson confidence intervals, and aggregate significance levels.
Table 10.
Sensitivity to the masking threshold at 20 dB.
Fig 18.
Sensitivity of PAPR reduction and reconstruction performance to the masking threshold at 20 dB.
Table 11.
High-order modulation experiment with modulation-specific PolyNet-Volterra training.
Fig 19.
High-order modulation performance after modulation-specific PolyNet-Volterra training.
Table 12.
Doppler robustness of PolyNet-Volterra at 20 dB.
Fig 20.
Doppler robustness of PolyNet-Volterra under increased maximum Doppler index.
Table 13.
Complexity and RTX 3090 GPU inference latency of neural reconstruction models.
Table 14.
Measured RTX 3090 GPU training time for the 150-epoch training schedule.
Table 15.
Full-frame input ablation at 20 dB.
Fig 21.
Full-frame input ablation comparing full-frame and masked-only neural reconstruction.