Process Characterization: Broadband Large-Signal Characterization and Modeling for Wireless System Integration

Introduction	Radio-frequency (RF) measurements are applied extensively in the deployment of commercial wireless communication systems. They are crucial to all stages of system development, from device modeling to circuit design and system performance characterization. NIST's RF and microwave measurement teams are addressing the critical need for accurate measurements of nonlinear electrical networks and supporting industrial standards development. Background/Impetus Customers Goals Impact SED Milestones R&D Team Achievements
Background/ Impetus	As wireless networks are pushed beyond the limits of network analysis, large signal descriptions are required to characterize devices. Our long term goal is to dramatically improve state of the art measurements and modeling of high-frequency wireless systems and components under large signal conditions. NIST has assembled a multidisciplinary team and established a new measurement facility for large signal device-level characterizations. Existing techniques are limited by measurement uncertainty in phase; we require additional external information to obtain a high frequency characterization. Moreover, general system level modeling techniques are inadequate and fail to exploit the wealth of data which is now available at the device level. We are developing a standard nonlinear device in order to reduce measurement uncertainty and developing and evaluating new models for nonlinear devices.
Customers	The customers for the broadband project are the U. S. electronics and communications industries.
Goals	The goals for the broadband project are: Develop and support general methods of characterizing nonlinear components, circuits, and systems used in digital wireless communications; and refine and transfer these methods through interactions with industrial research and development laboratories.
Impact	The broadband project will enable more efficient wireless system design and could impact every segment of the wireless community, both in existing commercial systems and higher-frequency links being considered.
	The milestones for the broadband project are:
FY02 Milestones	Develop metric-based diagnostics for quantifying the stability of the measurement system. Apply performance metrics to compare competing prediction models for various nonlinear devices, especially "standard" superconducting devices and example diode circuits. Begin development of semi-empirical statistical models for nonlinear verification devices in either the time or frequency domains.
FY01 Milestones	Design and analyze Nonlinear Network Measurement System (NNMS) measurement stability study. Develop performance metrics for comparing data from nonlinear circuit characterizations and models. Develop statistical methods for selecting NNMS large-signal operating conditions that do not exceed frequency-measurement constraints of the hardware.
R&D Team	Dom Vecchia, Statistical Engineering Division, ITL Kevin Coakley, Statistical Engineering Division, ITL Jolene Splett, Statistical Engineering Division, ITL Hari Iyer, Statistical Engineering Division, ITL Don DeGroot, Radio Frequency Technology Division, EEEL Jeff Jargon, Radio Frequency Technology Division, EEEL Kate Remley, Radio Frequency Technology Division, EEEL Jim Booth, Electromagmetic Technology Division, EEEL K. C. Gupta, University of Colorado Dominique Schreurs, ESAT-TELEMIC, Belgium
Achievements	The achievements of the broadband project include: We designed measurement repeatability studies and identified significant warm-up effects and more troubling trends in the experimental data collection system. We are investigating whether these systematic effects are connected to the device-under-test (DUT) or the measurement system itself. We developed a statistical procedure for detecting significant harmonics in the measured signals. Based on this procedure, we selected favorable experimental operating conditions for model development. At these conditions, there is no significant harmonic content beyond the capabilities of the measurement system. We developed of a general class of metrics for quantifying the relative performance of candidate prediction models. The metrics we are currently considering allow for prediction errors to be weighted differently by frequency. In general, the choice of the weights depends on the particular application and the particular device under test (DUT). We have shown that the equal-weight metric has a clear interpretation in the time domain. Demonstrated performance metric application by comparing predictions from a CAD simulation model to measured responses in the repeatability study. The analysis uncovered problems with data quality, and model validity and robustness. Demonstrated the application of performance metrics to interlaboratory comparisons by computing scores for power-bias sweeps on several devices measured on the NIST Nonlinear Network Measurement System and on a comparable Agilent system in Brussels, Belgium. Implemented automated data sweeps and designed repeatability studies on three nonlinear devices to be used in multi-lab comparison studies. Analyzed results for determination of nonlinear behavior needed in the selection of power-bias configurations suitable for development of new prediction models.

Date created: 2/6/2002
Last updated: 2/6/2002
Please email comments on this WWW page to sedwww@nist.gov.