Semantic Communication with Entropy-and-Channel-Adaptive Rate Control over Multi-User MIMO Fading Channels

Abstract: Although significant improvements in transmission efficiency have been achieved, existing semantic communication (SemCom) methods typically use a fixed transmission rate for varying channel conditions and transmission contents, leading to performance degradation under harsh channel conditions. To address these challenges, we propose a novel SemCom method for wireless image transmission that integrates entropy-andchannel-adaptive rate control mechanism, specifically designed for multi-user multiple-input multiple-output (MU-MIMO) fading channels. Unlike existing methods, our system dynamically adjusts transmission rates by leveraging the entropy of feature maps, channel state information (CSI), and signal-to-noise ratio (SNR), ensuring optimal communication resource usage. It incorporates feature map pruning, channel attention, spatial attention, and multi-head self-attention (MHSA) to effectively prioritize critical semantic features while minimizing unnecessary transmission overhead. Experimental results demonstrate that the proposed system outperforms separated source and channel coding and deep joint source and channel coding (Deep JSCC), in terms of rate-distortion performance, flexibility, and robustness, particularly in challenging scenarios such as low SNR, imperfect CSI, and inter-user interference.