Joint Coding and Modulation for Robust Semantic Communication in Satellite Communications

While satellite communication (SatCom) has demonstrated great potential in enhancing network coverage and connectivity, it still faces challenges such as poor channel quality and bandwidth limitations. Fortunately, semantic communication (SemCom) has been recognized as a promising solution to greatly reduce bandwidth consumption and enhance transmission robustness. However, applying SemCom to SatCom raises a critical concern of data security. Specifically, in SemCom, the Semantic Knowledge Base (SKB) typically requires frequent updates to adapt to changes in concepts. Nevertheless, due to the higher latency in SatCom, attackers might exploit this delay to inject adversarial examples, leading to semantic deviations in SKB. To adress above issues, we propose a new framework named JCM-SAM, which is based on SemCom Joint Coding-Modulation (JCM) assisted by the Spatial Attention Mechanism (SAM). This framework uses the SAM to improve the precision of semantic information and combines the SKB to recover original data information, thereby enhancing communication bandwidth. Furthermore, to maintain the consistency of information in SatCom, we incorporate adversarial training into the SKB. Simulation results demonstrate that the proposed method can significantly reduce bandwidth requirements and adapt to low Signal-to-Noise Ratio (SNR) conditions. Compared to baselines, it has achieved a 7.8% improvement in Peak Signal-to-Noise Ratio(PSNR). Moreover, in the Jacobian-based saliency map attack(JSMA), our method’s PSNR only drops by about 24.8%.