Title :
Advanced Neural Network Techniques for GaN-HEMT Dynamic Behavior Characterization
Author :
Orengo, G. ; Colantonio, P. ; Giannini, F. ; Pirola, M. ; Camarchia, V. ; Guerrieri, S. Donati
Author_Institution :
Dept. of Electron. Eng., Tor Vergata Univ., Roma
Abstract :
This paper presents a new approach to build RF dynamic behavioral models, based on time-delay neural networks (TDNNs), suitable for FET devices, and capable to identify the working class and to characterize both short- and long-term device memory, through a time-domain training procedure, for a wide range of input power levels. The presented model has been effectively applied to GaN-based devices, working in class A, AB and B
Keywords :
III-V semiconductors; gallium compounds; high electron mobility transistors; neural nets; semiconductor device models; wide band gap semiconductors; FET devices; GaN; GaN-HEMT dynamic behavior characterization; RF dynamic behavioral models; device memory; high electron mobility transistor; thermal model; time-delay neural networks; time-domain training procedure; Delay effects; Finite impulse response filter; Intrusion detection; Network synthesis; Neural networks; Neurons; Nonlinear filters; Radio frequency; Sampling methods; Time domain analysis; Dynamic behavioral model; GaN; Neural Network; Thermal model;
Conference_Titel :
European Microwave Integrated Circuits Conference, 2006. The 1st
Conference_Location :
Manchester
Print_ISBN :
2-9600551-8-7
DOI :
10.1109/EMICC.2006.282799
