IMS2026 Technical Program

In an increasingly congested spectrum landscape, companies, regulators, and policymakers are looking at new frequencies. With large chunks of untapped bandwidth, and the increasing maturity of the required technology, the sub-THz band offers significant promise for the wireless communications world. At the same time, existing services and stakeholders in the band, e.g., from the passive remote sensing and radio astronomy communities, need to be protected. Finally, international and national regulations limit emissions above 100 GHz largely based on considerations derived at lower frequency, overlooking the unique characteristics of electromagnetic wave propagation above 100 GHz, e.g., molecular absorption, and of the corresponding technology, e.g., the extreme directivity of the antennas. There is a growing need for 1) new propagation models and measurements across frequencies that capture the stakeholders’ diverse needs and ways of interacting with the spectrum; 2) new circuits, antenna designs, and interference cancellation techniques for sharing and coexistence; and 3) dialogue between the scientific and other stakeholders to understand and model Radio Frequency Interference. With this panel, we want to foster the dialogue between often siloed communities. To do so, we have invited representatives from the wireless communications, radioastronomy, and remote sensing community, including policy advocates and experts.

Recent advances in artificial intelligence (AI) and machine learning (ML) are transforming the way wireless components and complex electromagnetic (EM) systems are conceived, designed, and deployed. This session explores how ML-enabled optimization techniques are redefining applied electromagnetics, spanning the full pipeline from computational electromagnetics (CEM), uncertainty quantification (UQ), and antenna design to impactful applications such as magnetic resonance imaging (MRI), orthopaedic diagnostics, and remote sensing of snow and environmental parameters. By embedding AI and ML into EM modeling and optimization workflows, engineers can accelerate design cycles, navigate high-dimensional design spaces, and achieve performance levels that are difficult to reach with conventional approaches.
Beyond algorithms, the session emphasizes the critical role of data in driving the quality, robustness, and trustworthiness of AI-based solutions. High-fidelity simulation data, measurement-driven datasets, and hybrid physics-informed approaches are discussed as essential enablers for reliable learning and generalization. Attention is also given to the challenge of bridging ambition and deployment—moving AI-enhanced EM techniques from proof-of-concept demonstrations to deployable, validated systems operating under real-world constraints.