A novel distributed source coding model which named semantic-aware multi-terminal source coding problem is proposed and studied in the paper. This is motivated by the new communication paradigm being aware of semantic information, in which invisible semantic features are observed by multiple agents, and both semantic and observation reconstructions are imposed distortion constraints. The theoretical analysis of this model is provided in this work, in which we present a generalized Berger- Tung based sum rate region considering the semantic source, and further obtain upper and lower bounds when sources are joint Gaussian distributed. Under this case, the tradeoff between two distortions and optimal rate allocation scheme are discussed. Moreover, since the model couples the conventional multiterminal coding and CEO problems, the degeneration of generalized bounds to existing works are shown. Finally, we also present the sum rate bounds in special cases when sources are Bernoulli and distortion measure adopts logarithmic loss.

翻译：本文提出并研究了一种名为语义感知多终端信源编码问题的新型分布式信源编码模型。该模型受新兴的语义信息感知通信范式启发，其中不可见的语义特征由多个智能体观测，且语义重建与观测重建均受到失真约束的限定。本文对该模型进行了理论分析，提出了考虑语义源的广义Berger-Tung和速率区域，并进一步推导了当信源服从联合高斯分布时的上界与下界。在此条件下，讨论了两种失真之间的权衡及最优速率分配方案。此外，由于该模型耦合了传统多终端编码与CEO问题，本文展示了广义界向现有工作的退化过程。最后，针对信源为伯努利分布且失真度量采用对数损失的特定情形，给出了和速率界。