With the rapid evolution of communication technologies and the increasing complexity of wireless environments, semantic communication systems face significant challenges in maintaining performance under dynamic channel variations. Most existing deep joint source-channel coding (DJSCC) methods are designed under the assumption of stable transmission conditions, particularly a fixed or consistent signalto-noise ratio (SNR) between training and deployment. However, real-world wireless channels are highly dynamic due to multipath fading, interference, and environmental changes, leading to severe performance degradation when such assumptions are violated. To address this issue, this chapter introduces two novel channel adaptation mechanisms for semantic communication. First, an attention-based channel adaptation strategy is developed to dynamically adjust feature importance according to fluctuating SNR levels, improving the robustness of semantic representations. Second, a diffusion-based adaptive scheme is proposed, where channel noise is modeled as the forward process of a diffusion model, and reverse denoising is performed at the receiver to recover high-fidelity semantic features. Together, these mechanisms enhance the adaptability and reliability of semantic communication systems, bridging the gap between theoretical models and practical wireless deployments.