Kiran Ilyas
I would like to apply for the KIVI EngD Award 2026 in recognition of my EngD project, AI/ML-Based Nearfield to Far-field Mapping for mm-wave ICs in EMC Modelling, completed in June 2026 at Eindhoven
University of Technology in collaboration with NXP Semiconductors.
My motivation for undertaking this project was to address a practical engineering challenge in the
development of increasingly compact and complex millimeter-wave integrated circuits. As these
technologies become more prevalent in automotive radar and wireless communication systems,
predicting their electromagnetic radiation behavior becomes essential for electromagnetic
compatibility (EMC). Conventional approaches often involve computationally demanding simulations
and complex post-processing, limiting their efficiency during iterative design.
The main contribution of my EngD project was the development of a systematic, data-driven framework
connecting electromagnetic simulation, automated data generation, and artificial intelligence. Rather
than focusing solely on training a neural network, I approached the problem as an integrated
engineering workflow.
The project progressed through three complementary phases. First, I benchmarked full-wave
electromagnetic solvers to establish a suitable simulation foundation. Second, I independently
developed an automated workflow for generating consistent near-field and far-field datasets across
parameterized designs. Third, I independently implemented and evaluated deep learning architectures
capable of predicting far-field radiation characteristics directly from reactive near-field information.
This approach demonstrates how established electromagnetic modeling techniques can be combined
with machine learning to address a challenging engineering problem. The results showed that neural
networks can capture important radiation characteristics and generalize to previously unseen
geometries within the investigated design space, offering a promising route toward reducing repeated
computationally intensive simulations.
Beyond the technical results, independently developing the simulation and data-generation workflow
and the AI models required integrating expertise in electromagnetics, numerical simulation,
programming, and machine learning into a coherent engineering solution. Working within an industrial
project at NXP, I took responsibility for translating the technical challenge into a structured
development process, making engineering decisions, and assessing the resulting models against
practical requirements.
I believe this work reflects the purpose of the EngD programme: applying advanced technical
knowledge to develop innovative solutions for real-world engineering challenges. Its value lies not only
in the demonstrated near-field to far-field prediction capability but also in the reusable methodology for
creating simulation-based datasets and evaluating AI-assisted electromagnetic modeling approaches.
I welcome the opportunity to present my EngD project for consideration for the KIVI EngD Award 2026.
Through this application, I hope to share how combining electromagnetic simulation, automation, and
artificial intelligence can contribute to more efficient and scalable EMC-oriented design workflows,
while demonstrating the value of an industry-driven, multidisciplinary engineering approach

