基于神经网络的接收机宽带非线性行为建模

为了预测复杂电磁环境下接收机的非线性效应,文中基于实数时延径向基函数神经网络,构建了具有记忆效应的接收机非线性神经网络模型。分别采用K-均值聚类算法和正交最小二乘法对模型的隐含层中心和权值进行选取和学习,并用接收机的输入输出实测数据对模型进行训练。通过宽带信号的同相和正交两个分量对模型进行验证。模型仿真结果与实测数据相吻合,模型的归一化均方误差可达-41.88 dB。该结果表明,所构建的神经网络模型具有较快的收敛速度、良好的建模精度和泛化能力。     

一种精确的锁相环IP模块行为级建模

根据模拟硬件描述语言Verilog-A的特点,抽取欲设计锁相环各模Vegilog-A建立的相应模块的行为级模型中,并且根据晶体管级仿真结果对行为级模型中的参数进行实时修正,建立了比较精确的中心频率为100MHz的PLL行为级模型.     

Sigma Delta调制器高效行为级建模

提出一种宏模型和Verilog-A模型相结合的方法对两阶、1位量化的Sigma Delta调制器进行建模.对调制器中的关键模块采用宏模型建模,对功能性模块采用Verilog-A描述.在Cadence环境下,基于华虹NEC 0.25μmCMOS工艺对模块进行设计和仿真,并与实际电路模块仿真结果和仿真时间进行对比,给出两种情况下调制器总体电路的SNR仿真结果.结果显示:这种建模方法既达到了较高的精度,又取得了较快的仿真速度.     

非线性自动压扩的开关电流电路行为级建模方法

提出了一种新的基于非线性压扩函数自动构造的开关电流电路行为级建模方法,从而简化电路的建模和仿真.与原有的建模方法相比,该方法不仅可以对模型的误差分布进行有效地调控,而且能够降低模型的误差.为了验证本文所提出的行为级建模方法,对几种开关电流电路进行了建模和模拟试验.     

PLL-FS行为级建模仿真进行噪声和抖动性能预估的方法

本文提出了一种对锁相环频率合成器(PLL-FS)行为级建模后仿真,进行噪声和抖动性能分析的方法。新方法借鉴了最新的理论成果,结合工程实践,处于Top-Down设计流程的顶端。实例表明此方法可在PLL-FS设计之初对所设计系统的相位噪声和抖动性能有较精确的预估,并可据此调整设计参数,选择恰当的电路结构,从而显著提高了设计效率。     

非线性自动压扩的开关电流电路行为级建模方法

提出了一种新的基于非线性压扩函数自动构造的开关电流电路行为级建模方法,从而简化电路的建模和仿真.与原有的建模方法相比,该方法不仅可以对模型的误差分布进行有效地调控,而且能够降低模型的误差.为了验证本文所提出的行为级建模方法,对几种开关电流电路进行了建模和模拟试验.     

一种基于模糊神经网络的非线性系统模型辨识方法

该文提出一种非线性系统的模型辨识方法。通过结构的辨识(学习)和参数的辨识(学习),构造了一个模糊神经网络,经调整网络的权值,获得一个精确的模糊模型。对两个非线性系统辨识的仿真结果验证了该方法的有效性。     

一种频率合成器的建模优化与电路设计

根据数模混合集成电路系统级和行为级快速验证的需求,设计了一种卫星导航系统射频接收机前端的频率合成器。传统行为级模型一般是基于理想环路进行参数提取,误差较大。为此,首先,分别利用MATLAB和Verilog-AMS对频率合成器建立理想行为级模型与非理想行为级模型,并根据行为级模型提取与优化的环路参数,采用SMIC 180 nm CMOS工艺设计仿真电路级频率合成器;其次,建立MATLAB噪声模型,对电路级各个模块的噪声进行拟合,评估频率合成器系统的整体噪声性能。所提出的频率合成器设计方法对电路级设计具有前瞻性的指导,并有助于电路级的设计优化。     

迭代接收机性能建模方法研究进展

针对迭代接收机的性能建模方法,从等效模型、传递参数、性能提取及传递关系等4个方面概述了研究成果,并归纳了各自特点及发展趋势.服从高斯分布的对数似然比信息等效模型是迭代接收机性能建模研究的基础,以互信息为传递参数并通过仿真获得互信息传递关系图,以误比特率或保真度为传递参数的半公式化传递关系及以互信息为传递参数的半公式化传递关系分别是性能分析、性能优化及链路联合设计等3种应用场景下性能建模的代表性方法,是迭代接收机性能建模研究的发展方向.     

决策支持系统的建模方法研究

决策过程涉及的信息繁杂、对于模型要求比较高,尤其在处理非结构化的建模时,常用的数学方法效率就很低.介绍了系统辨识理论在构造决策支持系统模型的思想,该思想在非结构化和半结构化过程模型问题的处理中有较高的参考价值.     

A vector intermodulation analyzer applied to behavioral modeling of nonlinear amplifiers with memory

A large signal vector intermodulation network analyzer with a dynamic range of 90 dB and phase resolution of better than 2/spl deg/ is reported. The analyzer is used in conjunction with a multislice behavioral model to characterize memory effects in three different RF power amplifiers: an MOSFET instrumentation amplifier, a multistage GaAs/silicon-based broadband microwave integrated-circuit amplifier, and an SiGe HBT monolithic-microwave integrated-circuit amplifier. The multislice behavioral model architecture builds on conventional single-tone AM-AM and AM-PM modeling extended to capture long-term memory effects that are characterized by asymmetric intermodulation distortion (IMD). Phase asymmetries of upper and lower IMD are captured. A systematic procedure for extracting the model is presented.     

Volterra-mapping-based behavioral modeling of nonlinear circuits and systems for high frequencies

Presents and validates a discrete-time/frequency-domain approach to the problem of Volterra-series-based behavioral modeling for high-frequency systems. The proposed technique is based on the acquisition of samples of the input/output data, both of which are sampled at the Nyquist rate corresponding to the input signal. The method is capable of identifying the time-/frequency-domain Volterra kernels/transfer functions of arbitrary causal time-invariant weakly nonlinear circuits and systems operating at high frequencies subject to essentially a general random or multitone excitation. The validity and efficiency of the proposed modeling approach has been demonstrated by several examples in high-frequency applications and good agreement has been obtained between results calculated using the proposed model and results measured or simulated with commercial simulation tools.     

Switching flow-graph nonlinear modeling method formultistate-switching converters

Switching power converters operating in a multistate switching mode (more than two states) feature multidimensional control over their state variables. In this paper, a large-signal multistate modeling method is developed based on the switching flow-graph method to study the steady-state and dynamic properties of pulse-width-modulated (PWM) multistate-switching power converters for the continuous conduction mode. This modeling method translates a switching power converter directly to its graphic dynamic model and uses graphical representation to reveal the cause and effect relationship of the dynamics within a multidimensional power converter. A three-state buck-boost circuit is conceived with two duty ratios controlling two outputs as an example to test this modeling method. Experimental results confirm the theoretical prediction. This multistate-switching flow-graph modeling method is very general, easy to use and accurate, and it provides deep physical insight for engineering design     

On the use of behavioral modeling within the RFIC design flow: Satellite receiver case study

In order to generate a first-time right design, system level modeling and simulation is a major step. Thus, performance evaluation of integrated wireless systems requires the development of RF behavioral models compatible with the microelectronic design tools. Firstly, this paper shows the need of such models within the RFIC Top–Down design flow. Then, behavioral modeling techniques are presented and several classical RF block models are described. Those models are used within an RFIC satellite receiver validation example and finally, several performance evaluation examples allow demonstrating the capabilities of behavioral modeling, compared to other very used modeling techniques as mathematical and baseband approaches.     

Analog behavioral modeling by multi-companding and wavelet collocation method

In this paper, we develop a wavelet collocation method with multi-companding for behavioral modeling of analog circuits. In the multi-companding procedure, the nonlinear companding algorithm is developed to control the error distribution continuously, while the adaptive scheme is employed to reduce the number of used wavelets. Consequently, the proposed multi-companding algorithm can not only modify the modeling error distribution continuously but also decrease the number of basis functions efficiently. Moreover, the companding function generation is automatic and can be applied for the behavioral modeling of any analog circuits. Jun Tao received the B.S degree in electrical engineering from Fudan University, China, in 2002. Now she is currently working toward the Ph.D. degree in micro-electronic engineering at the Fudan University. Her research interest includes analog behavioral modeling, analog circuit simulation and DFM. Xuan Zeng (M97) received the B.Sc. and Ph.D. degrees in electrical engineering from Fudan University, Shanghai, China, in 1991 and 1997, respectively. She joined the Electrical Engineering Department, Fudan University in 1997 and became a full professor in Microelectronics Department in 2001. Now she serves as the Vice Director of ASIC & System State key Lab. and the Associate Head of Microelectronics Department Fudan University. She was a visiting professor in the Electrical Engineering Department, Texas A&M University, USA and Microelectronics Department of TU Delft, Netherland in 2002 and 2003 respectively. Her research interests include DFM, analog and mixed signal design automation (behavioral modeling, circuit simulation and analog layout generation), high speed interconnect analysis and design and ASIC design. Dr. Zeng received the Cross-Century Outstanding Scholar Award from the Ministry of Education of China in 2002. She was selected into “IT Top 10” in Shanghai China in 2003. She served in the technical program committee of IEEE/ACM ASP-DAC in 2000 and 2005. Dian Zhou received the B.S degree in physics and M.S degree in electrical engineering from Fudan University, China, in 1982 and 1985, respectively, and the Ph.D. degree in electrical and computer engineering from the University of Illinois in 1990. He joined the University of North Carolina at Charlotte as an assistant professor in 1990, where he became an associate professor in 1995. He joined the University of Texas at Dallas as a full professor in 1999, and joined Fudan university as a Changjiang Professor in 2003 (on-leave from the University of Texas at Dallas). Currently, he serves as the dean of Microelectronics School, director of National Key Lab. on ASICs and Systems, and director of Miro-nano-electronics Innovation Platform at Fudan University. His research interests include: High-speed VLSI systems, CAD tools, mixed-signal ICs, and algorithms. Charles Chiang received his Bachelor degrees from the Department of Political Science, Tunghai University at Taichung, Taiwan in 1980, and Department of Computer Science, New Mexico State University, Las Cruces, New Mexico in 1986. Then he had his Masters and Ph.D. degree from the Department of Electrical Engineering and Computer Science, Northwestern University, Illinois in 1988 and 1991, respectively. After working at IBM and EDA companies for 10 years, he joined the Advanced Technology Group at Synopsys, Inc. in 2001. His research interests include routing, placement, floorplan, and signal integrity. His main research focus is now on design for manufacturability (DFM). Dr. Chiang has been a Senior Member of IEEE since 1998. He received the Superior Design Recognition award and the ADAL award from IBM Rochester in 1993 and 1994, respectively. He is one of the top 15 winners with new patent filing in 2005 and 2006 in Synopsys. He has served on the technical committee of ICCAD from 2004 to 2006, on that of Field Programming Logic (FPL) from 2002 to 2003, as well as on the committee of ASP-DAC in 2007. He has published more than 40 technical papers and filed 10 US patents.     

Optical receiver front-end nonlinear distortion

The harmonic modulation products of a PIN-GaAs MESFET optical receiver front end are found to be about 20 dB smaller than those of a PIN-bipolar transistor front end, using the `two-frequency optical signal method?. It is pointed out that the design parameters in the PIN-GaAs MESFET front end, optimised from the high-sensitivity viewpoint, are compatible with the requirements for low distortion in analogue transmission systems.     

Large-signal behavioral modeling of nonlinear amplifiers based on load-pull AM-AM and AM-PM measurements

This paper presents an improved behavioral modeling technique that generates large-signal models for nonlinear amplifiers or devices based on load-pull AM-AM and AM-PM measurement datasets. The generated behavioral model characterizes the incident and scattering waveforms at two ports in the frequency domain based on the large-signal scattering function theory. The advantage of this technique is that it is derived entirely from load-pull measurements and provides an analytic method to utilize the load-pull measurements in practical designs. Examples are given to demonstrate the ability of the behavioral models to predict the load-related nonlinearities of the device-under-test.     

Time-domain envelope measurements for characterization and behavioral modeling of nonlinear devices with memory

This paper presents a calibrated four-channel measurement system for the characterization of nonlinear RF devices such as power amplifiers. The main goal of this study is to perform the characterization of the bandpass response of a nonlinear device-under-test (DUT) driven by modulated carriers. The proposed setup enables the generation of L- or S-band (1-4 GHz) carriers with a modulation bandwidth up to 100 MHz. The carrier harmonics generated by the nonlinear DUT are ignored and considered to be sufficiently filtered. This characterization setup enables calibrated time-domain measurements of the complex envelopes of both incoming and outgoing RF waves at the input and output of the DUT. This means that the fundamental and harmonic frequencies of the envelope are measured and processed. A large set of modulation formats can be generated by using a computer-controlled arbitrary waveform generator. Complex envelopes are measured by using a four-channel sampling scope. The proposed calibrated setup can be used to study or to validate linearization techniques of power amplifiers. This characterization tool is also well suited for the extraction and validation of behavioral bilateral models of nonlinear RF analog equipment exhibiting memory effects.     

A novel radiometer receiver stabilization method

A novel approach to solving the temperature stabilization problem of radiometer receivers is discussed, in which the classical temperature stabilization is replaced by a software-based compensation for any local temperature deviations observed. It is shown that a tunnel diode has to be incorporated as a square-law detector in order to validate this new approach. The resulting measurement uncertainty is calculated and an approximation of the effect of aging is given.     

Macromodeling of nonlinear transistor-level receiver circuits

In this paper, a modeling methodology for macromodeling transistor-level receiver circuits has been presented. A few receiver modeling techniques have been proposed in the past, but these modeling techniques only address the loading effect of the receiver circuits, i.e., the input characteristics of the receivers. In this paper, a modeling methodology that addresses both the loading effect as well as the output characteristics of the receiver has been proposed. This modeling technique is simple, accurate, and has huge computational speed-up over transistor-level receiver circuits. To model the input characteristics of the receiver, spline function with finite time difference (SFWFTD) and recurrent neural network (RNN) modeling methods have been used. The output characteristics of the receiver are modeled using a combination of receiver static characteristics and a delay element that takes into account the timing delay of the receiver. The accuracy of the modeling approach has been tested on some test cases and results show good accuracy and substantial speed-up compare to transistor-level receiver circuits. The proposed modeling technique has been extended to multiple ports to estimate sensitive effects like simultaneous switching noise (SSN) when multiple receivers are switching.