Engineering the environment using electromagnetic surfaces is one possible way of improving the performance of mm-Wave band deployments for 5g.
Engineered electromagnetic surfaces (EES) are surfaces which can be fabricated on a thin layer of plastic using a low-cost printing technology.
These surfaces can be installed at strategic locations in indoor and outdoor propagation environments to improve the coverage or mitigate interference at millimeter-wave bands for 5G applications.
EES Element Samples
The computational burden of the genetic algorithm can be intractable for fine discretizations of the EES element, requiring smart heuristics to aid in the search. This is where we believe machine learning may be a game changer.
In this project, we set out to understand the capabilities of Generative Adversarial Networks (cGANs) and conditional convolutional GANs (ccGAN) in the generation of electromagnetic engineered surfaces (EES).
Our results indicate conditional GANs can potentially be used to aid the design of Electromagnetic Engineered Surfaces (EES), improving the accuracy by at least 3 fold compared to a random generation process.