An Explainable Deep Semantic Coding for Binary-Classification- Oriented Communication

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(9), P. 4608 - 4608

Published: April 22, 2025

Semantic communication is emerging as a promising paradigm, where semantic coding plays an essential role by explicitly extracting task-critical information. Prior efforts toward often rely on learning-based feature extraction methods but tend to overlook data compression and lack rigorous theoretical foundation. To address these limitations, this paper proposes novel explainable deep framework, considering binary mixed source classification task at the receiver. From information-theoretic perspective, we formulate problem that jointly optimizes rate accuracy subject distortion constraints. solve problem, leverage learning techniques variational approximation develop practical schemes. Experiments CelebA dataset CIFAR-10 demonstrate proposed schemes effectively balance accuracy, which aligns with formulation.

Language: Английский

Estimation of Carrier Frequency Offset Plagued by IQ Mismatch Using Least-Squares Interpolation of the DFT Coefficients

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 107505 - 107515

Published: Jan. 1, 2024

The rapid development of wireless communication systems demands extremely high reliability in transmission links. However, non-ideal characteristics RF links, such as carrier frequency offset (CFO) and IQ mismatch (IQM), can lead to impairments, severely affecting system performance. Advanced signal processing techniques are indispensable mitigate the Within processing, synchronization issues, including CFO estimation schemes for synchronization, should be prioritized. traditional susceptible interference from harmonics caused by IQM, resulting severe performance degradation. Therefore, based on least squares (LS) interpolation method, we develop a scheme robust IQM using $2l _{0}+1$ ( notation="LaTeX">$l _{0}\geqslant ~3$ ) discrete Fourier transform (DFT) coefficients. proposed estimator combines coarse fine estimation. In first stage, peak position amplitude spectrum is detected provide rough second stage (fine estimation), an LS equation relationship between observation vector matrix established with observed DFT samples rotation factors. precise extracted principle equation, utilizing sample points. Because structure considered method combat IQI. Test results verification platform indicate that improves accuracy at 20 dB input power −45 dBm. Accuracy 9 −65 Complexity analysis shows increases most 10.74% compared LS-based methods.

Language: Английский

A semantic communication model for the task of high quality image transmission to edge-end devices

Internet of Things, Journal Year: 2024, Volume and Issue: unknown, P. 101384 - 101384

Published: Sept. 1, 2024

Language: Английский