Theory Of Electromagnetic Intelligent Agents With Applications To Mimo And Doa Systems

We propose the concept of electromagnetic intelligent agents (EM-IA) as a general framework for a new generation of antenna systems concurrently employing electromagnetics, communication theory, and artificial intelligence within one integrated platform. The proposed strategy emphasizes certain machine learning aspects that can be gleaned from communication-system-theoretic models carefully constructed using the antenna current Green's function formalism when applied to Tx/Rx antennas and the environments inside which they live. The system intelligence feature is enacted through reprogrammable artificial neural networks attached to the antennas' terminals and trained by data generated through the electromagnetic model. The proposed EM-IA scheme is demonstrated with a concrete example geared toward adaptive mutual coupling compensation in receiving MIMO and DoA array systems operating within dynamic environments.