IEEE MTT-S Young Professionals Workshop on Modeling and Optimization for Active Devices

Co-chairs:
Gian Piero Gibiino (University of Bologna, Italy), Justin King (Trinity College Dublin, Ireland)

Abstract:
This virtual workshop is aimed at discussing recent and prospective trends in the broad field of RF active device modeling, from devices to systems. Two sessions of presentations from Young Professionals (YPs) speakers will introduce their ongoing research work in the context of transistor and RF amplifier modeling, including optimization-based machine learning techniques. Field experts will provide their takeaways in a scientific panel discussion, as well as their specific tips to YPs in a career advice platform.

Panelists
Prof. J. Bandler (McMaster University, Canada)
Prof. P. L. Gilabert (Universitat Politècnica de Catalunya, Spain)
Prof. S. Khandelwal (Macquarie University, Australia)
Prof. J. C. Pedro (University of Aveiro, Portugal)
Prof. M. Pirola (Politecnico di Torino, Italy)
Prof. M. Rudolph (Brandenburg University of Technology, Germany)
Prof. Q. J. Zhang (Carleton University, Canada)

Schedule
Tuesday, 25th October 2022, 10:00 – 16:15 UTC (11:00 – 17:15 London/Dublin Time, 12:00 – 18:15 Paris/Rome Time)
This workshop will be conducted virtually via Zoom. It is open to everyone free of charge. Please register in advance to receive the Zoom link. 

Workshop Agenda
10:00 – 10:05 UTC Welcome address and intro to Session 1

10:05 – 10:35 UTC [S1-1] Petros Beleniotis, Brandenburg University of Technology, Germany, “Modeling Dispersive Effects of GaN HEMTs for Microwave PA Design”
Abstract – The talk provides a modeling strategy for dispersive effects in microwave PA design based on GaN HEMT technology. Commonly used dopants in the buffer of GaN HEMT and native defects in III-N semiconductors enhance trapping effects and often result in a significant mismatch between HEMT’s behavior and models. The presentation consists of an extended analysis of a trap description for the compact model ASM-HEMT and its extraction procedure. The proposed model simulates the drain lag and employs a parameter scaling technique to improve model accuracy. Measurement and simulation examples comparing a static model with the drain-lag one will be presented and discussed.

10:35 – 11:05 UTC [S1-2] Luís Cótimos Nunes, University of Aveiro, Portugal, “Characterization and Modeling of Slow Dynamics of AlGaN/GaN HEMTs and their Impact on RFPA Linearization
Abstract – Deep-level traps evidenced by AlGaN/GaN HEMT devices, with their inherent slow transients, are known to be very detrimental to the PAs’ linearizability. This work addresses the challenge of characterizing the slow trapping transients observed in these devices, at guaranteed invariant thermal dissipation conditions. Starting by understanding the physical mechanism behind the trapping effects, and how their transients are strongly dependent on the temperature, the talk then moves on to an extensive characterization and modeling of these effects. Finally, the talk demonstrates the importance of the coupled electrothermal and trapping effects, showing their impact on the GaN/HEMT-based PAs’ linearizability.

11:05 – 11:35 UTC [S1-3] Sachin Yadav, imec, Belgium, “Substrate Effects in GaN-on-Si HEMT Technology for RF FEM Applications”
Abstract – A cost-effective integration of GaN HEMTs on large area (>200 mm) silicon substrates not only benefits from standard CMOS back-end-of-the-line processing but also the wafer-level co-integration with Si-CMOS, enabling complex functionality and better performance than the standalone counterparts. However, for advanced wireless communication applications (e.g. 5G), GaN-on-Si HEMTs present thermal management and RF substrate loss related challenges. In this presentation, we will discuss the methodology to study and model RF substrate losses and non-linearities for GaN-on-Si substrates. Using the developed methodology, the impact of GaN HEMT integration on substrate RF performance is studied in a varying bias-temperature space.

11:35 – 12:05 UTC [S1-4] Nicholas C. Miller, Air Force Research Laboratory, USA, “Nonlinear RF modeling of GaN HEMTs with Fermi kinetics transport and the ASM-HEMT compact model”
Abstract – Rapid design and prototyping of next-generation microwave and mm-wave GaN technology requires reliable and accurate models. A paramount component of enabling first-pass design success of GaN microwave power amplifiers is the ability to model process variations inherent in GaN foundries. This talk will focus on the recent developments of connecting nonlinear ASM-HEMT models with the AFRL TCAD solver called Fermi kinetics transport (FKT). The FKT simulations will enable exact control of process variations in the GaN HEMT and will shed light on statistical nonlinear modeling of GaN microwave technology.

12:05 – 13:35 UTC Panel Session

13:35 – 14:05 UTC Career Advice Platform

14:05 – 14:10 UTC Intro to Session 2

14:10 – 14:40 UTC [S2-1] Xiaoqiang Tang, Hangzhou Dianzi University, China “Relating behavioural transistor device models to PA design”
Abstract – Behavioral modeling is used for fast and accurate characterisation of radio frequency (RF) power transistors. It is a principal technology for enabling ‘first-pass’ design success in electronic design automation (EDA). In particular, it is used for RF power amplifier (PA) design when short design time is of the utmost importance. This talk presents a study relating device modeling and PA design and attempts to strengthen the overall connection between the two, leading to an improved modeling framework for PA design.

14:40 – 15:10 UTC [S2-2] Lida Kouhalvandi, Dogus University, Turkey, “Intelligent-based Optimization Methods Applied for Designing High Performance Power Amplifiers”
Abstract – In the fifth-generation technology, high power amplifiers (HPAs) are the essential devices and for meeting the desired specifications, strong optimization methods are required. Electronic design automation (EDA) tools as AWR, ADS, etc. are useful for optimizing radio frequency designs typically with less than 20 parameters. However, when the dimension of variables is growing, these tools are not successful enough and intelligent-based methods are required. This presentation devotes to present various intelligence-based optimization methods employed for designing and optimizing HPAs, lead to have high performance outcomes in terms of output power, gain, efficiency, and linearity.

15:10 – 15:40 UTC [S2-3] Mattia Mengozzi, University of Bologna, Italy, “Machine Learning Methods for the Linearization and Global Optimization of Multiple-Input RF Power Amplifiers”
Abstract – The talk will introduce the main challenges posed by modern multi-input multiple-output (MIMO) architectures for high-efficiency Power Amplifiers (PA), such as supply- and load-modulated PAs, as well as for MIMO PAs for beamforming arrays. The talk will describe generalized digital predistortion (DPD) models and algorithms to improve the global behavior of multi-input PAs by exploiting the additional degrees of freedom of these topologies. Multi-objective optimization algorithms and surrogate-based techniques as well as other methods derived from machine learning will also be introduced.

15:40 – 16:10 UTC [S2-4] Wantao Li, Universitat Politècnica de Catalunya, Spain, “Digital linearization techniques for multiple-input, single output power amplifiers”
Abstract – The talk will focus on the digital baseband signal processing, including but not limited to linearization, to guarantee linearity specifications in high efficient amplification topologies based on dynamic load or dynamic supply strategies that include multiple-input, single output (MISO) power amplifiers. In this context, the suitability, advantages, drawbacks and challenges of some of the machine learning solutions proposed in the field of DPD linearization will be discussed.

16:10 – 16:15 UTC Closing remarks