Semantic Pilot Signal Enabled Adaptive Image Transmission in Semantic Communications

Semantic communications have demonstrated significant advantages in image transmission compared to conventional bit-based communication systems. Most existing semantic communication systems for image transmission focus only on the average reconstruction quality rather than the reconstruction quality of individual images. Moreover, these systems require prior knowledge of the signal-to-noise ratio (SNR) at both the transmitter and receiver for SNR-adaption, limiting their practical applicability in dynamic environments. To address the aforementioned problems, this paper proposes a Semantic Pilot Signal enabled Semantic Communication (SPS-SC) framework. In the framework, the semantic pilot signals are dynamically selected by semantic classification of images, enabling the receiver to quantify channel-induced semantic distortion via pilot signal analysis. By combining this distortion assessment with user-defined reconstruction quality requirements, the receiver adaptively determines the optimal compression ratio via codebook lookup, which is then fed back to the transmitter to achieve efficient image transmission. Simulation results demonstrate that the proposed SPS-SC framework is of outstanding effectiveness and robustness under varying channel conditions.