Grant: Strategic Projects - Group

The proposal is built on a recent breakthrough in high-frequency computer-aided design, namely neuro-space mapping. Combining the powers of neural network modeling and space mapping optimization, neuro-space mapping opens the door for a larger and more fundamental technology. It is becoming a logical answer to the growing challenges in modeling and design optimization of high-frequency effects in RF/microwave and millimeter-wave components and multi-gigabit electronic packages in communication networks.The proposed research will create the critical mass for the emerging neuro-space mapping technology. Exploiting innovative mappings between electromagnetic/physics-based modeling and fast/approximate modeling, we aim to achieve design solutions at much higher levels of accuracy, speed and scale than possible today. The research will unify a wide spectrum of modeling techniques from general black-box to accurate application-specific modeling, from data-based to analytical/compact modeling, and from coarse to fine modeling. Comprehensive frameworks will be developed to serve as the foundation for a new era of computer-aided modeling and design, including yield/manufacturability-driven design optimization with EM/physics accuracy at circuit and system levels, a task beyond the reach of today's technology.Our work will establish a new state of the art in mathematically- and industrially-validated engineering modeling and optimization. By addressing high-frequency challenges, it contributes to pushing bandwidth limits in communication systems. The new computer-aided modeling in place of conventional trial-and-error modeling accelerates the application of new and emerging devices in the design of communication systems. The work will be carried out in close interaction with Canadian industry, strengthening Canadian competitiveness in next generation communication networks.