Simulation, modeling and optimization of devices, circuits and systems

点击翻译按钮获取中文摘要

First-pass success and time-to-market are crucial for manufacturability-driven design of telecommunication circuits.  This necessitates powerful, accurate and fast computer-aided design (CAD) tools for component modeling, circuit simulation and automated design optimization.  Simulations based directly on physics and electromagnetics (EM) are traditionally computationally intensive, but offer high accuracy, handle arbitrary geometrical shapes and are valid to millimeterwave frequencies.  Their direct exploitation in optimization, statistical design and design with tolerances remains challenging. Pioneering work in design centering, tolerance and statistical design for manufacturability, tuning and yield-driven optimization (IEEE MTT-Society 2004 Microwave Application Award) will be broadened into a flexible theoretical and software framework for computer-aided engineering of wireless, radio-frequency (RF) and microwave components and circuits.  The framework will be based on our novel space-mapping optimization technology, which we pioneered in 1993 and which has already resulted in diverse and significant modeling and design successes in the engineering community in Canada and internationally.  New algorithms are expected to deliver in a handful of fine or high-fidelity model evaluations the accuracy expected from classical direct optimization using sequential linear programming. We will continue to work with Dr. Kaj Madsen (Technical University Denmark) on the convergence theory for space-mapping algorithms.  This should help engineering practitioners in developing or discriminating between coarse engineering models.  We will develop prototype software engines which exploit both full-wave EM simulators and fast, empirical, coarse or surrogate device models.  We will continue to explore links with artificial neural network technology for device modeling, e.g., neuro space mapping (Q.J. Zhang).  We will develop efficient, robust algorithms for EM-based optimization exploiting space mapping, available adjoint sensitivities, surrogate models (John Dennis) and the trust region methodology.