In the sub 100-nm regime, MOSFET parameter extraction has become a challenging task. Commonly used gradient based methods have many difficulties such as good initial guess requirement, singularities in objective functions, etc. Genetic Algorithm (GA), which does not suffer from these problems, is reported for parameter extraction of various state-of-the-art MOSFET models recently. Compared to GA, the Particle Swarm Optimization (PSO) algorithm  is reported to be more efficient for several applications and is also shown to perform better for MOSFET parameter extraction . In this paper, a novel “memory loss (ML)” feature is introduced in the PSO algorithm for the first time, which further improves its efficiency.
Journal: TechConnect Briefs
Volume: 3, Nanotechnology 2008: Microsystems, Photonics, Sensors, Fluidics, Modeling, and Simulation – Technical Proceedings of the 2008 NSTI Nanotechnology Conference and Trade Show, Volume 3
Published: June 1, 2008
Pages: 845 - 848
Industry sector: Sensors, MEMS, Electronics
Topics: Compact Modeling