Abstract
The rapid expansion of Internet of Things (IoT) devices places growing pressure on limited spectrum resources. Semantic communication offers a promising solution by transmitting only the essential meaning of data, thereby reducing redundancy. However, the dynamic and heterogeneous nature of IoT environments leads to fluctuating channel conditions that can degrade semantic reliability. To address this challenge, we propose semantic Hybrid Automatic Repeat reQuest (sem-HARQ), a framework that integrates Hybrid Automatic Repeat reQuest (HARQ) with Joint Source–Channel Coding (JSCC) to enable reliable semantic transmission. To mitigate the adverse effects of multipath fading and noise, we introduce a semantic equalizer module that leverages conditional generative modeling to correct semantic distortions. Furthermore, three sem-HARQ schemes are developed: sem-HARQ-I, sem-HARQ with Weighted Combining (sem-HARQ-WC), and sem-HARQ with Synonymous Combining (sem-HARQ-SC). While sem-HARQ-I and sem-HARQ-WC retransmit the same semantic signals, sem-HARQ-SC introduces new semantic redundancies via synonymous mapping across HARQ rounds. At the receiver, sem-HARQ-WC and sem-HARQ-SC enhance reliability by fusing current and previous semantic features, effectively reusing information from prior erroneous transmissions. Simulation results demonstrate that the proposed sem-HARQ framework significantly improves semantic reliability compared to no-HARQ baselines. Among the three designs, sem-HARQ-SC achieves the best performance, albeit with higher complexity, while sem-HARQ-I provides the simplest but least effective solution.
Funding Information
  • Scientific Research Innovation Capability Support Project for Young Faculty (SRICSPYFBS2025134)
  • National Natural Science Foundation of China (62171200)
  • Fundamental Research Funds for the Central Universities (21625361)
  • Science and Technology Development Fund, Macau Special Administrative Region, China (0020/2025/RIB1)
  • Team-based Research Fund (TBRF/2024/1.10)
  • Research Matching Grant (CP/2025/1.1)
  • Research Grants Council (RGC) of Hong Kong Special Administrative Region, China (UGC/FDS16/E06/25)

This publication has 40 references indexed in Scilit: