By Topic

Circuits and Systems, 1992. ISCAS '92. Proceedings., 1992 IEEE International Symposium on

Date 10-13 May 1992

Go

Filter Results

Displaying Results 1 - 25 of 103
  • A self-organizing neural network for nonlinear filtering

    Page(s): 2629 - 2632 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (312 KB)  

    A neural network based on the combination of a feature map and linear filters is proposed as a generalized adaptive processor for multidimensional nonlinear mapping. The self-organizing part of the system provides a progressively finer embedding of the input space as more units are added to the network. The linear filters, which tap from the memory, provide the function approximations. Learning is achieved with simple rules of the Hebb's type with no backpropagation needed. Some preliminary results on two-dimensional patterns show the potential of this approach View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A general purpose VLSI chip for robot axis motion controller

    Page(s): 3005 - 3008 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (444 KB)  

    A VLSI-based realization and implementation of an entire robot axis controller is proposed as a single-chip joint controller. This chip requires a 68000-bus family microprocessor (MP) and a power transistor module. It forms a three-component complete robot servo-level controller which correlates all joint electromechanical activities with the supervisor level (main robot CPU). The proposed chip is equipped with an on-chip RS485 port for communication with the CPU. It is also provided with a pin-to-pin interface to its accompanied MP. All joint digital and analog feedback signals connect directly to the chip View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Homotopy methods for solving decoupled power flow equations

    Page(s): 2833 - 2839 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (580 KB)  

    Explores the application of probability-one homotopy algorithms for solving the load flow equations under the decoupling assumption. The existence of solutions and convergence properties of homotopy methods are interpreted within the power systems framework, via an analogy between nonlinear resistive circuits in steady state and power systems. These ideas are further related to system stability. The effectiveness and performance of various homotopy functions are presented through simulation results for several small examples of power systems View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Feedforward neural network for handwritten character recognition

    Page(s): 2884 - 2887 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (340 KB)  

    An analysis of feedforward neural networks for handwritten character recognition was performed to improve the learning capability and accuracy of classification, which are limiting factors of back-propagation. The authors describe two methods which attempt to tackle the shortcomings of back-propagation yet keep the feedforward organization of the neural network. These methods give results comparable to back-propagation, while requiring less training time and a simpler architecture. The first method rejects any pattern which differs from the training data more than a threshold, established during training. The second method involves clustering techniques selecting the most representative patterns as cluster centers. Both methods present the design of a neural network for handwritten digit recognition, and are based on the Parzen window estimates defining the vector space for different classes View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • High-precision design techniques in BiCMOS

    Page(s): 2703 - 2706 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (176 KB)  

    The author discusses the utility of bipolar devices in high-accuracy, BiCMOS-based circuit design. Examples are given which exploit the greater stability, lower 1/f noise, and higher operating voltage of bipolar devices. The examples show that BiCMOS circuits easily extend the performance boundaries normally reached in CMOS-only design. Compared to pure CMOS, BiCMOS adds flexibility to high-precision circuit design, resulting in performance advances with shorter development time and lower risk View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A sigma-delta modulation based analog adaptive filter

    Page(s): 2657 - 2660 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (320 KB)  

    In analog implementation, analog multiplier accuracy, circuit complexity, and lack of a lossless analog delay line have made it difficult to implement least-mean-square (LMS) analog adaptive filter structures. The authors propose an analog adaptive filter structure based on pulse density modulation to circumvent these problems by using a digital delay line and an analog tap value calculation of the adaptive algorithm. Sigma-delta modulation is used as an efficient conversion scheme to the binary representation, and the binary stream is processed using analog techniques. The adaptive filter can be realized using switched capacitor integrators for filter coefficients and only 1-b word length for the delay line. The high updating frequency of the filter taps leads to fast adaptation View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Wave-domino logic: timing analysis and applications

    Page(s): 2949 - 2952 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (308 KB)  

    Wave-pipelining has been investigated extensively as a design method to increase clock frequencies of digital systems. The authors explore wave-pipelining of domino CMOS circuits because of their compact layouts and shorter delay time. A timing model and constraints are given for deriving the minimal period. Partitioning and restructuring of circuits for wave-pipelining are discussed. A test chip has been designed using a novel form of domino-logic to verify the concept View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Power system small-disturbance stability analysis using circuit-theoretic techniques

    Page(s): 2829 - 2832 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (272 KB)  

    The authors review power system small-disturbance (SD) stability using circuit theoretic techniques. SD stability is the stability of an equilibrium operating point under small perturbations such as load fluctuations. Different system models are considered for frequency and for voltage problems. The SD stability analysis is usually performed via linearization of the system equations around the equilibrium of interest. An equilibrium is SD stable if all eigenvalues of the corresponding Jacobian matrix lie in the open left half plane. Necessary and/or sufficient conditions for SD stability are often obtained through positive-definiteness and M-matrix properties of a certain matrix. These properties are derived from the node admittance matrix of a network of positive resistances View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Image restoration using a multilayer perceptron with a multilevel sigmoidal function

    Page(s): 2917 - 2920 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (308 KB)  

    The problem of restoring a blurred and noisy image having many gray levels, without any knowledge of the blurring function and the statistics of the additive noise is considered. A multilevel sigmoidal function is used as the node linearity. Restoration is achieved by exploiting the generalization capabilities of the multilayer perceptron network. To overcome the burden of training time a segmentation scheme is suggested. Simulation results are also provided View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Electrical power network decomposition for parallel computations

    Page(s): 2761 - 2764 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (364 KB)  

    Introduces an electrical power network decomposition methodology for the network equation solution by block-iterative methods implemented on parallel computers. The proposed methodology is based on the building up of a fixed number of subnetworks by aggregating network nodes to a given number of so-called seed nodes. The aggregation method is controlled by a node ranking criterion based on the weights assigned to the network nodes reflecting the magnitude of their electrical coupling. The resulting decompositions are usually conducive to well-conditioned iterative processes, with acceptable computational load balancing, and relatively low communication requirements for the parallel solution of the network equations. Results of computational tests performed on medium-size real networks are presented View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A novel neural network for object recognition with blurred shapes

    Page(s): 2901 - 2904 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (360 KB)  

    Presents a neural network approach to shape recognition. The emphasis is on the development of an effective representation method to increase the degree of robustness in recognition of shapes which may be blurred by noise. A space-perturbation neural network is described which is characterized by two important properties. (1) The network can be trained using error back-propagation with only noise-free data, which avoids the convergence problem due to possibly large variance in each shape class. (2) The space-perturbation arrangement enables the network to discover class features independent of random variations in shape and hence not blurred by random variations in the input. As a recognition task, the classification of four closed planar shapes was chosen. For comparison, the neural network classifier based upon the extended training technique was implemented to perform the same task. Performance evaluation results for 4000 testing shapes from various blur conditions are given View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Simple algorithms for tracing solution curves

    Page(s): 2801 - 2804 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (364 KB)  

    The author presents some simple and practical algorithms for tracing implicitly defined solution curves. These algorithms use hyperspheres instead of hyperplanes that are used in the typical predictor-corrector algorithms. Effective techniques for preventing the reversion phenomenon of the curve tracing are also proposed. The proposed algorithms are geometrically clear and can be easily programmed. It is also shown that the algorithms can be easily implemented on existing circuit simulators such as SPICE View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Multivariate positive functions under coefficient perturbation

    Page(s): 2733 - 2736 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (312 KB)  

    Testing of positive realness of a rational function when its coefficients vary over given intervals is required in practical applications such as design of adaptive control systems. The k-variate positive function is first related to a (k+1)-variate Hurwitz polynomial. Then, using the Kharitonov-type results for Hurwitz polynomials, similar results are derived for multivariate positive functions and multivariate very strict positive functions. This derivation shows that for a k-variate interval positive function, only 8×2k extremum positive functions need be tested. This is against the previously reported 16×4k functions for the multivariate case and 32 for the single-variate case. The interval positive real functions require the testing of 4×2k extremum functions View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Investigation of transistor networks concerning feedback structures

    Page(s): 2868 - 2871 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (264 KB)  

    Gives a computational realization of a theorem of N.O. Nielsen and A.N. Willson, Jr. (1980) for studying the DC operational point of transistor networks in a qualitative manner. Although this theorem is constructive in nature a direct approach leads to tedious combinatorial computations. Based on a backtracking technique an algorithm is proposed which circumvents these difficulties. Some heuristics are explained. By means of some examples the usefulness as well as the limitations of the algorithm are shown View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • An analysis of inductive peaking in high-frequency amplifiers

    Page(s): 2848 - 2851 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (308 KB)  

    Presents a theoretical treatment of inductive peaking. Although the analysis is general, it is specifically targeted for the design of monolithic broadband transimpedance amplifiers in photoreceivers. The authors derive closed-form, symbolic solutions for the small-signal transimpedance and analyze the effects of different types of inductive peaking on system performance. Based on these closed-form solutions, it is possible to optimize the amplifier bandwidth by proper choice of inductance. A comparison between the results of the symbolic analysis and microwave SPICE simulation data is shown and demonstrates a remarkably good match View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Interval trigonometric polynomials and wavenumber response of n -D FIR digital filters

    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (36 KB)  

    Summary form only given. The scope for determining the lower and upper bounds for the wavenumber response of a multidimensional interval finite impulse response (FIR) filter from the corresponding bounds for each element of a subset of extreme filters is investigated View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Direct conversion using lowpass sigma-delta modulation

    Page(s): 2653 - 2656 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (288 KB)  

    A method for quadrature demodulation by subsampling with sigma-delta analog-to-digital converters is discussed, and a correlator receiver structure based on this method is described. The almost completely digital demodulator structure is based on second-order sampling and sigma-delta analog-to-digital converters. A theoretical performance analysis is presented, and a measurement system for the method is described. The proposed structure can be implemented on a single integrated circuit View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Adaptive estimation of frequency in power networks

    Page(s): 2741 - 2744 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (272 KB)  

    A common problem in power networks is the estimation of the fundamental frequency of harmonic signals. The authors propose and examine two related methods that use the autoregression as an underlying model, avoid the estimation of its parameters, and determine the fundamental frequency directly. They are intended to operate on signals which, besides the fundamental, may contain harmonic components. In addition, they should operate in high signal-to-noise ratio environments. They are adaptive in nature and do not require heavy computations. The methods are parametric and are based on the fact that n real sinusoids can be represented by a linear difference equation with constant coefficients. The order of the equation of 2n and its coefficients are characterized by a special symmetry. Simulation results are provided that demonstrate the performance of the methods View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Micro-statistic LMS filtering

    Page(s): 2613 - 2616 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (344 KB)  

    The theory for the class of microstatistic least-mean-square (LMS) filters which requires the adaptation of the statistical characterization of the set of decomposed signals is developed. The authors develop the theoretical framework for adaptive microstatistic filters for applications where the second-order statistics of the threshold signals are not known, or when they may be nonstationary. A multilevel threshold decomposition is used such that real valued stochastic processes can be filtered and the computation complexity of the algorithm can be arbitrary reduced. The superiority of the new adaptive algorithms is shown analytically as well as by way of simulations View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Design automation technology for codesign: status and directions

    Page(s): 2669 - 2672 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (236 KB)  

    Computer-aided design tools and design automation for hardware/software codesign have developed to the point that commercial tools supporting codesign activities are under development or in use. The current state of codesign technology and specific needs in codesign tools and environments in the future are discussed. The architecture design and assessment system (ADAS) developed at Research Triangle Institute was the first implementation reported in the literature that supported the concurrent or cooperative design of hardware and software. The ADAS model is the basic approach used to date by codesign tools. Hardware synthesis systems, modeling paradigms, design metrics, and scaling-up techniques are discussed View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Network theoretic conditions for existence and uniqueness of steady state solutions to electric power circuits

    Page(s): 2821 - 2828 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (628 KB)  

    Surveys available network theoretic results on conditions under which the unique and operationally acceptable load flow solution exists. The main topic is the problem of establishing system-dependent bounds on input and network topology within which the unique solution exists. Both decoupled real-power-phase angle and reactive-power-voltage problems are studied. The particular emphasis is on the conservativeness of the sufficient conditions under which the power network has the properties of a nonlinear resistive network. Several new results and open questions are reported regarding conservativeness of sufficient conditions for solution existence View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A neural-based digital communication receiver for inter-symbol interference and white Gaussian noise channels

    Page(s): 2933 - 2936 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (356 KB)  

    A multilayer perceptron with the extended Kalman filter (EKF) training algorithm is investigated as a communication receiver when intersymbol interference and white Gaussian noise are present. Besides discussing the complexity of the EKF algorithm, it is shown that the EKF has better performance than the conventional backpropagation (BP) training algorithm in the sense that is requires less training steps and also results in proper training even when the BP did not. With 2500 training symbols, the EKF resulted in about 4-dB performance improvement over the conventional BP View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Neural networks for computed tomography

    Page(s): 2893 - 2896 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (272 KB)  

    Proposes a novel application of neural networks to cognitive tasks of computed tomography (CT). The principle of Hopfield type neural networks to reconstruct the image from the projected densities is described. The technique of reconstruction is based on the algebraic reconstruction technique (ART). Simulation results for a model of an image, where the reconstruction space was divided into 32×32 elements with 9 color degrees, show the satisfactory performance in terms of accuracy and computation time by application of the neural networks View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Shift invariant pattern recognition by associative memory

    Page(s): 2913 - 2916 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (412 KB)  

    Presents a network architecture that can perform shift invariant pattern recognition. The network is composed of a preprocessing block and an associative memory block which is a recurrent neural network. The preprocessing network is designed in such a way that the output of this block is almost invariant under shifted input patterns. The associative memory block is employed to recall the original patterns stored in the system. A step by step design procedure for realizing shift invariant pattern recognition is provided View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Recent results in robust parametric stability and control

    Page(s): 2707 - 2711 vol.6
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (348 KB)  

    The author describes some recent results obtained on the robust stability and control of systems containing parametric uncertainty. He begins with Kharitonov's theorem on interval systems and its generalization to control systems containing linearly correlated interval parametric uncertainty, obtained by H. Chapellat and S.P. Bhattacharyya in 1989. The relationship of this result, called the linear CB theorem and of its generalization to the multilinear case, to extremal stability subsets and other connections constitutes an extensive and coherent theory of robust parametric stability. This theory is summarized View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.