Electromagnetic information theory (EIT) is an emerging interdisciplinary subject that integrates classical Maxwell electromagnetics and Shannon information theory. The goal of EIT is to uncover the information transmission mechanisms from an electromagnetic (EM) perspective in wireless systems. Existing works on EIT are mainly focused on the analysis of degrees-of-freedom (DoF), system capacity, and characteristics of the electromagnetic channel. However, these works do not clarify how EIT can improve wireless communication systems. To answer this question, in this paper, we provide a novel demonstration of the application of EIT. By integrating EM knowledge into the classical MMSE channel estimator, we observe for the first time that EIT is capable of improving the channel estimation performace. Specifically, the EM knowledge is first encoded into a spatio-temporal correlation function (STCF), which we term as the EM kernel. This EM kernel plays the role of side information to the channel estimator. Since the EM kernel takes the form of Gaussian processes (GP), we propose the EIT-based Gaussian process regression (EIT-GPR) to derive the channel estimations. In addition, since the EM kernel allows parameter tuning, we propose EM kernel learning to fit the EM kernel to channel observations. Simulation results show that the application of EIT to the channel estimator enables it to outperform traditional isotropic MMSE algorithm, thus proving the practical values of EIT.

翻译：电磁信息理论（EIT）是一门新兴交叉学科，融合了经典麦克斯韦电磁学与香农信息论。EIT的目标是从电磁视角揭示无线系统中的信息传输机制。现有关于EIT的研究主要集中于自由度分析、系统容量及电磁信道特性，但尚未阐明EIT如何改善无线通信系统。为回答这一问题，本文首次对EIT的应用进行了创新性论证。通过将电磁知识融入经典MMSE信道估计器，我们首次观察到EIT能够提升信道估计性能。具体而言，首先将电磁知识编码为时空相关函数（STCF），称为电磁核（EM kernel）。该电磁核作为信道估计器的先验信息。由于电磁核采用高斯过程（GP）形式，我们提出了基于EIT的高斯过程回归（EIT-GPR）方法进行信道估计。此外，针对电磁核的参数可调性，我们提出电磁核学习算法，使其适配信道观测数据。仿真结果表明，将EIT应用于信道估计器可使其性能优于传统各向同性MMSE算法，从而验证了EIT的实用价值。