We work on fundamental and applied aspects of Computational Electromagnetics (CEM),  scientific machine learning, uncertainty quantification and optimization techniques, to enable advances in current, emerging and future wireless communication components (antennas, reconfigurable intelligent surfaces) and systems (5/6 G and beyond). 

In this page, you will find information about recent news and publications, our research areas, our published work and pre-prints, our group members, and Prof. Sarris' teaching activities

Research Highlights

ACES2021-1313.mp4

Generalizable Neural Network Models for Wireless Propagation Modeling

We have introduced neural network based propagation models that can generalize with respect to geometry, position of transmitter and receiver and operating frequency of indoor/outdoor channels (i.e. they can rapidly solve new geometries, with transmitters operating from new positions and new frequencies, well beyond those included in the training set). 

Multiphysics Modeling with Physics-Informed Neural Networks

We demonstrated a new paradigm for solving multiphysics problems utilizing robust, single-physics techniques (such as FDTD or finite elements) combined with neural networks to model associated physics (such as thermal effects).

Propagation modeling with ray-tracing and the vector parabolic equation method (VPE)

Accurate modeling of radiowave propagation at 2.4 GHz along 3 km of tunnels and open-air sections in the London (UK) underground (measurements provided by Thales Canada/UK). Details for this work are provided in: N. Sood et al, "Integrating Physics-Based Wireless Propagation Models and Network Protocol Design for Train Communication Systems," in IEEE Transactions on Antennas and Propagation, Dec. 2018.

Shape optimization for metasurface unit cells with B-splines and DGTD simulations (IMS2021 Student Paper Competition Finalist)

Research by Qiming Zhao