.png){kind=icon}
.png){kind=icon}
Return
Deep Learning-Based Bit-Wise Flexible Error Protection Modulation
.png){kind=icon}
PRE
AI
DOI：10.1109/LCOMM.2025.3646846.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This letter proposes a novel deep learning–based scheme for bit-wise flexible error protection (FEP) at the modulation level, which employs a custom composite loss with tunable protection weights and dynamic weighting coefficients to enable the same autoencoder architecture to flexibly realize both equal error protection (EEP) and unequal error protection (UEP). Moreover, a newly designed loss function, used within this composite loss, explicitly penalizes decoding errors with large Hamming distances, thereby suppressing their prediction probabilities and directly improving bit error rate (BER) performance. Simulation results demonstrate that the proposed FEP modulation scheme, enhanced by the newly designed loss function, achieves better BER performance than the conventional loss function, realizes distinct BER differentiation under UEP, achieves identical BER under EEP, and adapts effectively to diverse protection granularities and channel conditions.
Keywords:
Deep learning
autoencoder
flexible error protection
equal error protection
unequal error protection
Journal
IF:
4.4
Papers:
1.3W
Citations:
2.2W