Causal and Passive Parameterization of S-Parameters Using Neural Networks

Date

2020-10-01

Author

Torun, Hakki Mert
Durgun, Ahmet Cemal
Aygun, Kemal
Swaminathan, Madhavan

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

214
views

0
downloads

Neural networks (NNs) are widely used to create parametric models of S-parameters for various components in electronic systems. The focus of deriving these models has so far been numerical error reduction between the NN-generated S-parameters and the data source. However, this is not sufficient when creating such NNs since it does not guarantee predicted S-parameters to be physically consistent, i.e., passive and causal, which restricts their use cases. This article, therefore, proposes a causality enforcement layer (CEL) and passivity enforcement layer (PEL) that can be used in NNs, which ensures that NN-predicted S-parameters are of a passive and causal system. To achieve this, we utilize Kramers-Kronig relations and singular value properties of S-parameters during the training stage with the purpose of learning a physically consistent representation. This enables end-to-end training where no postprocessing is required to ensure physical consistency. We demonstrate the effectiveness of the presented approach for three different design applications, where the goal is to predict S-parameters from dc to 100 GHz. The results show that when NNs are trained using CEL and PEL, the predicted S-parameters are characterized as 100.0% causal and passive while having the same level of accuracy as NNs that solely focus on error minimization.

Subject Keywords

Electrical and Electronic Engineering, Radiation, Condensed Matter Physics

URI

https://hdl.handle.net/11511/69939

Journal

IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES

DOI

https://doi.org/10.1109/tmtt.2020.3011449

Collections

Department of Electrical and Electronics Engineering, Article

Suggestions

OpenMETU
Core

Transformation Electromagnetics Based Analysis of Waveguides With Random Rough or Periodic Grooved Surfaces Ozgun, Ozlem; Kuzuoğlu, Mustafa (Institute of Electrical and Electronics Engineers (IEEE), 2013-02-01) A computational model is introduced which employs transformation-based media to increase the computational performance of finite methods (such as finite element or finite difference methods) for analyzing waveguides with grooves or rough surfaces. Random behavior of the roughness is taken into account by utilizing the Monte Carlo technique, which is based on a set of random rough surfaces generated from Gaussian distribution. The main objective of the proposed approach is to create a single mesh, and to ana...
Mutual coupling of printed elements on a cylindrically layered structure using closed-form Green's functions Acar, R. C.; Dural, G. (EMW Publishing, 2008-01-01) A hybrid method to calculate mutual coupling of electric or magnetic current elements on a cylindrically layered structure using closed-form Green's functions is presented. When rho = rho' and phi is not very close to phi', closed-form Green's functions are employed in the calculation of MoM matrix entries. When both rho = rho' and phi = phi', series representation of the spectral domain Green's functions do not converge, therefore closed-form Green's functions can not be employed. In that case MoM matrix e...
Hybrid connection of RF MEMS and SMT components in an impedance tuner Unlu, Mehmet; Topalli, Kagan; Atasoy, Halil Ibrahim; Demir, Şimşek; Aydın Çivi, Hatice Özlem; Akın, Tayfun (Elsevier BV, 2010-01-01) This paper presents a systematic construction of a model for a hybrid connected RF MEMS and SMT components in a reconfigurable impedance tuner. The double stub hybrid impedance tuner which employs a high number of MEMS switches is selected to demonstrate the feasibility of the connections. In the hybrid tuner, MEMS switches are actuated with DC bias signals, where SMT resistors de-couple RF from the DC lines. The hybrid tuner is realized in two steps, where the MEMS impedance tuner is fabricated on a glass ...
Modelling and noise analysis of closed-loop capacitive sigma-delta mems accelerometer Boğa, Biter; Külah, Haluk; Department of Electrical and Electronics Engineering (2009) This thesis presents a detailed SIMULINK model for a conventional capacitive Σ-Δ accelerometer system consisting of a MEMS accelerometer, closed-loop readout electronics, and signal processing units (e.g. decimation filters). By using this model, it is possible to estimate the performance of the full accelerometer system including individual noise components, operation range, open loop sensitivity, scale factor, etc. The developed model has been verified through test results using a capacitive MEMS accelero...
RC Performance Evaluation of Interconnect Architecture Options Beyond the 10-nm Logic Node Kıncal, Serkan; Abraham, Mathew C.; Schuegraf, Klaus (Institute of Electrical and Electronics Engineers (IEEE), 2014-06-01) This paper summarizes the findings of an RC performance modeling approach for evaluating various material and architecture options by which interconnect wires are incorporated onto integrated circuits. For the present dual-damascene structure, the grain boundary and surface scattering modes are identified as the top contributors to resistance degradation, along with the cross-sectional area consumed by the liner/barrier layers. Self-forming barriers, a technology that provides direct Cu-insulator interfaces...

Citation Formats

H. M. Torun, A. C. Durgun, K. Aygun, and M. Swaminathan, “Causal and Passive Parameterization of S-Parameters Using Neural Networks,” IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, pp. 4290–4304, 2020, Accessed: 00, 2021. [Online]. Available: https://hdl.handle.net/11511/69939.