News
Current News and Recent Work
Prof. Sarris undergraduate lectures on Electricity and Magnetism available online
Prof. Sarris lectures on the 2nd year undergraduate course on Electricity and Magnetism (ECE221) have been recorded and made available by the Educational Technology Office of the Faculty of Applied Science and Engineering here.
Congratulations to
Aris Seretis on receiving his Ph.D., successfully defending his thesis entitled "Physics-based Propagation Models enabled by Machine Learning".
Shutong Qi, on receiving an IEEE MTT-S Graduate Fellowship.
Qiming Zhao on receiving his Ph.D., successfully defending his thesis entitled "Adaptive Modeling of Electromagnetic Structures with DGTD and Subcell FDTD Methods" and starting a new career with Flexcompute.
Prof. Sarris selected as IEEE Antennas and Propagation Society Distinguished Lecturer
Prof. Sarris has been selected as a Distinguished Lecturer of the IEEE Antennas and Propagation Society for 2024-2026. The titles of his talks are:
The transformative impact of machine learning enabled computational electromagnetics on the future of wireless (general audience)
Scientific machine learning for electromagnetic field computations
From propagation models to physics-based digital twins of emerging wireless communication systems
Realistic propagation models for RIS-enabled communication channels
Efficient FDTD-based modeling of finite periodic structures
More information on the IEEE AP-S Distinguished Lecturer Program can be found here.
Comprehensive FDTD review paper published on Nature Reviews Methods Primers
A review of the FDTD method including its current state, recent developments, challenges and future has been published in Nature Reviews Methods Primers. The paper has been written by a team of authors including Profs. J. Simpson, F. Teixeira, J.P. Berenger, W. Chew, M. Okoniewski and C. Sarris.
Reconfigurable Intelligent Surface (RIS) modeling work presented at special session of the IEEE Int. Symp. on Antennas and Propagation in Portland, OR.
Our presence at the 2023 IEEE Int. Symposium on Antennas and Propagation included papers on robust propagation models for reconfigurable intelligent surfaces, benchmarking physics-informed neural networks for time-domain computational electromagnetics and generalizable neural network propagation models.
Prof. Sarris speaks at AI/ML bootcamp at IMS 2023 in San Diego, CA
Prof. Q.J. Zhang (Carleton University), Ulf Gustavsson (Ericsson) and Prof. Sarris held an AI/ML bootcamp at the 2023 IEEE MTT-S Int. Microwave Symposium in San Diego, CA, attracting 100+ attendees. The bootcamp was part of an array of AI/ML focused activities at the IMS, including three technical sessions and a panel session.
Paper by Aris Seretis et al. on ANN models for tunnel propagation wins the 2021 Premium Award for Best Paper in IET Microwaves, Antennas & Propagation
The award was given to the paper Artificial neural network models for radiowave propagation in tunnels, vol. , 14, Sept. 2020. Premium Awards are given by the IET to recognise the best research papers published during the last two years.
News Archive
ECE Teaching Award to Prof. Sarris for third year electromagnetics course
Prof. Sarris is a recipient of an ECE Departmental Teaching Award for his work in ECE320: Fields and Waves (a third year course on electromagnetic fields), in Fall 2021. Awards are decided by confidential vote of the students.
CEMpact at the 2022 European Conference on Antennas and Propagation (Madrid, Spain)
CEMpact will be present at the 2022 EuCAP in Madrid, with the following papers:
A Hybrid Machine Learning-Based Model for Indoor Propagation (Authors: Aristeidis Seretis and Costas Sarris) @Session CS14: AI for antennas and propagation: current trends and emerging applications
Abstract: A common limitation among many applications involving machine learning techniques is the availability of training data. In propagation modeling scenarios, measurement campaigns are usually undertaken for network planning decisions. However, this can be a challenging task, especially in electrically large environments. In these cases, simulation data generated by physics-based methods, such as ray tracing, can replace or augment the measured data. This paper provides a case study in a typical office environment, where both measured and simulated data are used to separately train two machine learning models. A hybrid model combines the predictions of these two models, to predict the signal levels at any location in the environment. In cases where the volume of measured data is insufficient, the hybrid model is shown to improve the accuracy of the overall predictions. It is also shown that a small number of measurements can improve the accuracy of a solver-trained model.
Convex Optimization of Reactively Loaded Antenna Arrays with Backlobe and Sidelobe Constraints (Authors: Michel A Nyffenegger ; Costas Sarris ; Hans-Dieter Lang) @Session E04: Optimization and machine learning in antenna design
Abstract: A semidefinite relaxation-based optimization framework originally developed for wireless power transfer systems in the near-field is adapted to maximize the antenna gain of reactively loaded antenna arrays in the far-field while simultaneously limiting the backlobe and sidelobe levels. The method is applied to a five-element linear dipole array and is successfully verified by simulation and measurements. The presented framework can be used for other types of loaded antennas as well as for geometrical optimization of dipole arrays.
New paper on scientific machine learning for FDTD computations
Preprint is available here
This paper demonstrates a deep learning based methodology for the rapid simulation of planar microwave circuits based on their layouts. We train convolutional neural networks to compute the scattering parameters of general, two- port circuits consisting of a metallization layer printed on a grounded dielectric substrate, by processing the metallization pattern along with the thickness and dielectric permittivity of the substrate. We also build a hybrid neural network including recursive neural network modules, to compensate numerical dispersion errors in coarse-grid FDTD. This novel dispersion compensation scheme allows us to generate accurate FDTD training data from fast, coarse-grid simulations.
Prof. Sarris re-appointed Editor-in-Chief of IEEE J-MMCT
Prof. Sarris has been appointed to a second term (2022-2024) as Editor-in-Chief of the IEEE Journal on Multiphysics/Multiscale Computational Techniques, a joint publication of the IEEE Microwave Theory and Techniques, Antennas and Propagation and Electromagnetic Compatibility Societies.
Undergraduate Research Featured at the 2021 IEEE International Symposium on Antennas and Propagation
4th year research project by G. Yang, S. Luo, S. Ji and D. Cui on "Physics-Informed Machine Learning Models for Indoor Wi-Fi Access Point Placement" was presented at the flagship IEEE conference on antennas/propagation.
Review paper on Machine Learning Models for Radiowave Propagation
The paper "An Overview of Machine Learning Techniques for Radiowave Propagation Modeling" (A. Seretis, C. Sarris) is available on IEEEexplore. Arxiv preprint is available here.
ML-based propagation modeling research featured in June 2022 special issue of the IEEE Transactions on Antennas and Propagation
Two papers from our group are included in the Special Issue of the IEEE Transactions on Antennas and Propagation, dedicated to "Artificial Intelligence in Radio Propagation for Communications" (June 2022).