This paper provides a framework for the analysis of error rate performance of maximal ratio combining multichannel reception of coherent linearly modulated systems over frequency-selective fading channels. Expressions for the error probability are developed, and the special cases for which closed-form or acceptable efficient numerical solutions are possible are delineated. A new efficient and accurate recursive method for the Ricean fading main resolved path case is presented for coherent quadrature amplitude modulation constellations. The effect of automatic gain control error on analysis of modulation schemes is demonstrated. Finally, numerical examples based on the maritime high-data-rate channel model are provided to demonstrate the usefulness of the framework.

In this paper we use a segmented Chebyshev orthongonal polynomial expansion approach to approximate amplitude and phase response for RLC interconnect simulation. This method has been shown to be efficient in reducing frequency sampling point number or expansion order. Experiments also show that the number of sampling points is not necessarily dependent on the interconnect circuit size, which makes possible large network simulation. Additionally, because of the smoothing-effect of the atan function in evaluation, the phase response, which has major impact on the transient edge of output response, can be easily approximated. Experiments show the proposed simulation approach achieves 30-50 times speed-up over spice3f4. 1.

We present a new realizable reduced order modeling technique for RLC interconnect trees. Both lumped and distributed wire models can be used with this technique. Provable stability is achieved by using Hurwitz polynomials. Moment computation process is avoided but moments can still be matched implicitly. In experiments, the proposed Hurwitz three-pole model can accurately and efficiently capture inductive effect for both near end and far end nodes.

A new category of moments-Amplitude and Phase moments (AP moments) are introduced for RLC interconnect delay estimation. We show that there are tight relationships between AP moments, circuit moments and central moments. The first order AP moment represents the Elmore delay while the higher order AP moments can be used to represent the error between the Elmore delay and the exact 50% delay from the view of gain and phase-shift variation. With the help of the physical meaning revealed by the AP moments, a closed-form 50% delay model-AP delay model is proposed for RLC interconnect delay estimation in terms of the first four AP moments. We also propose a new two-pole model (AP two-pole model) by matching the first two phase moments of the transfer function. The AP two-pole model can be used for more generally timing parameters estimation. The input signal's impact on delay estimation can be incorporated into these two delay models by simply combining the input signal's AP moments with the transfer function's AP moments. In our experiments these two models show significant accuracy improvement over the Elmore delay model.

This paper presents a recursive method for evaluating the error rate performance of multichannel reception of coherent MPSK systems operating over slowly Ricean fading channels. Each of the multiple channels is assumed to be a bandlimited two-sided discrete path model with complex Gaussian taps. The main path is assumed to be non-zero mean, corresponding to Ricean fading. Maximal ratio combining in the receiver is assumed, and an expression for the received phase density function is developed. A recursive method for solving the resulting error probability expression is then presented. This method is similar to those used for recursive computation of the Marcum Q function, and is both efficient and accurate. The method can be applied for both the flat and frequency selective Ricean fading cases. Examples based on a ship to ship high-data-rate line-of-sight channel model are provided to demonstrate the method.

Two-dimensional (2-D) adaptive filtering is a technique that can be applied to many image processing applications. This paper will focus on the development of an improved 2-D adaptive lattice algorithm (2-D AL) and its application to the removal of correlated clutter to enhance the detectability of small objects in images. The two improvements proposed here are increased flexibility in the calculation of the reflection coefficients and a 2-D method to update the correlations used in the 2-D AL algorithm. The 2-D AL algorithm is shown to predict correlated clutter in image data and the resulting filter is compared with an ideal Wiener-Hopf filter. The results of the clutter removal will be compared to previously published ones for a 2-D least mean square (LMS) algorithm. 2-D AL is better able to predict spatially varying clutter than the 2-D LMS algorithm, since it converges faster to new image properties. Examples of these improvements are shown for a spatially varying 2-D sinusoid in white noise and simulated clouds. The 2-D LMS and 2-D AL algorithms are also shown to enhance a mammogram image for the detection of small microcalcifications and stellate lesions.

The paper focuses on the development of an improved 2D adaptive lattice algorithm (2D-AL) and its application to the removal of correlated clutter to enhance the detectability of small objects in images. The two improvements proposed are increased flexibility in the calculation of the reflection coefficients and a 2D method to update the correlations used in the 2D-AL algorithm. The results of the clutter removal is compared to previously published ones for a 2D LMS (TDLMS) algorithm. 2D-AL is better able to predict spatially varying clutter than the TDLMS algorithm since it converges faster to new image properties. Examples of these improvements are shown. The TDLMS and 2D-AL algorithms are also shown to enhance a mammogram images for the detection of small microcalcifications.>

It has been proposed that CDMA systems can be assigned to spectral bands which are presently occupied by narrowband users to further increase spectral capacity. Such CDMA overlay systems could provide new options for efficient utilization of the spectrum with minimal disruption to existing narrowband users, especially if adaptive interference suppression techniques are utilized in the spread spectrum receiver. Previous studies have defined the SNR improvement ratio which can be achieved for tone interferers and for narrowband interferers for which the center frequency of the interference is at the carrier frequency of the CDMA signal. In this paper the bit-error-rate (BER) performance of the mobile-to-base link of a CDMA system for a single narrowband user which occupies a significant portion of the CDMA bandwidth is evaluated. It is shown that the narrowband model used in previous studies does not apply in this case, especially for the large, effective, bandwidths which are characteristic of the interferers in the overlay system. The dependence of the BER on the filter order, the bandwidth of the interference, and its center frequency relative to the CDMA carrier frequency are defined. Additionally the increase in BER for a digital implementation of the adaptive suppression filter relative to the optimal Wiener filter is characterized with respect to the adaptive time constant and the quantization errors due to finite wordlength. It is shown that these implementation errors can be made negligible compared to the errors which are characteristic of the optimal Wiener filter. Analytic results are validated by simulation for typical system parameters.>

We propose a novel paradigm for two-way circuit partitioning which minimizes the clock cycle. The replication technique is suggested for feedback loops to minimize the impacts of intermodule delays and the crossing edges when necessary. A flow timing cut is devised to produce partitions which can be guaranteed to achieve clock cycles equal to their lower bound with respect to the partitions using retiming. When the clock cycle optimization is the major objective and feedback loop sizes are not large, we propose an efficient, easy to implement algorithm which still guarantees achieving the lower bound clock cycle with respect to its partition. Experimental results have shown that our algorithms can achieve an average of 15% clock cycle time reduction compared to the best retimed results produced by 20 runs on each test case using a Fiduccia-Mattheyses algorithm.

This work studies the performance of dimensional least mean square (TDLMS) adaptive filters as prewhitening filters for the detection of small objects in image data. The object of interest is assumed to have a very small spatial spread and is obscured by correlated clutter of much larger spatial extent. The correlated clutter is predicted and subtracted from the input signal, leaving components of the spatially small signal in the residual output. The receiver operating characteristics of a detection system augmented by a TDLMS prewhitening filter are plotted using Monte-Carlo techniques. It is shown that such a detector has better operating characteristics than a conventional matched filter in the presence of correlated clutter. For very low signal-to-background ratios, TDLMS-based detection systems show a considerable reduction in the number of false alarms. The output energy in both the residual and prediction channels of such filters is shown to be dependent on the correlation length of the various components in the input signal. False alarm reduction and detection gains obtained by using this detection scheme on thermal infrared sensor data with known object positions is presented.

The authors compare the performance of the recursive least squares lattice (RLSL) and the normalized step-size stochastic gradient lattice (SGL) algorithms to that of the least mean square (LMS) transversal algorithm for the cancellation of sinusoidal interference. It is found that adaptive lattice filters possess a number of advantages over the LMS transversal filter, making them the preferred adaptive noise cancellation (ANC) filter structure if their increased computational costs can be tolerated. These advantages include faster convergence, notch bandwidths which are independent of the input power, and the generation of less harmonic distortion. The notch bandwidth of each filter structure is related to its respective convergence time constant. Experimental results obtained with the 32-b floating-point lattice development system (LDS) are presented to verify the analytical results.>

A recursive detection scheme for point targets in infrared images is described. Estimation of the background noise is done using a weighted autocorrelation matrix update method and the detection statistic is calculated using a recursive technique. A weighting factor allows the algorithm to have finite memory and deal with nonstationary noise characteristics. The detection statistic is created by using a matched filter for colored noise, using the estimated noise autocorrelation matrix. The relationship between the weighting factor, the nonstationarity of the noise and the probability of detection is described. Some results on one- and two-dimensional infrared images are presented.>

The authors study the behavior of the partial correlation (PARCOR) coefficients of the stochastic gradient adaptive lattice filter in response to a complex linear chirp FM signal in white Gaussian noise. A single-stage model for the behavior of the coefficients is developed, and the recovery error is derived. Also, an accurate model for a two-stage filter is derived.>

The performance of adaptive lattice filters in finite precision real-time hardware is largely unknown because of the complexity of implementing lattice algorithms. The authors show how arithmetic effects can deleteriously manifest themselves in both the stochastic gradient and the least squares adaptive lattice filters. Computer simulations and experimental results are used to verify an analytical analysis describing implementation effects of adaptive lattice filters. It is found that the performance of adaptive lattice filters can be substantially improved by manipulating the time recursive coefficients using direct updating techniques. Results are compared to those obtained from LMS (least mean square) transversal adaptive filters.>

The use of adaptive filters for the enhancement of images is studied. In particular, the enhancement of images where the region of interest has a small spatial extent compared to the noise is considered. A two stage approach for enhancing the desired signal is presented. At each stage, a two dimensional adaptive filter employing the least mean square (LMS) algorithm is used. By properly choosing the adaptation step size, in the first stage the colored noise is whitened, and in the second stage the desired signal is recovered from the white noise. Procedures for selecting the step size are discussed. Computer simulations to support the results are also presented.>

To make the application of the finite-element method practical in semiconductor device simulation, the authors have applied the Scharfetter-Gummel (S-G) scheme in conjunction with an accurate seven-point Gaussian quadrature rule to the assembly of the finite-element stiffness matrices and the right-hand-side vector of the semiconductor equations. The key of this method lies in accurate interpolation rules, which are derived on the basis of simple device physics considerations. The inherent simplicity and flexibility in the finite element formulation make the new method applicable to multidimensional problems. The simplicity of embedding the S-G scheme in the quadrature of finite-element assembly lends itself to all kinds of finite-element methods using various elements, shape functions, and weightings. The resultant exponential functional fitting avoids high discretization errors usually incurred by the classical finite-element discretization method. Solutions with high accuracy, even on coarse mesh, and a significant speed-up of convergence rate are obtained.>

POSM (pseudo-objective function substitution method), is an unconstrained optimization algorithm. This algorithm, requiring neither derivative calculation nor linear search step, substitutes the objective function by a second-order approximated formulation, enhancing the convergence rate substantially and implemented in the general-purpose Analysis and Design Program for Integrated Circuits, (ADIC-2.C). To show the effectiveness and efficiency of this algorithm, it was evaluated using a number of numerical examples and some practical IC designs.>

This paper presents the extension of bit-serial architectures for next-generation signal processing chips with floating-point and fault-tolerant capabilities. The architecture and design of a novel bit-serial floating-point complex multiplier-accumulator (CMAC) hardware is described. Fault-tolerant features are incorporated by embedding these CMACs into dynamically reconfigurable systolic arrays. The arrays can be utilized for FIR filtering, convolution, correlation and other signal processing applications based upon multiply-accumulate. Furthermore, these computational cores can be incorporated into more complex processors which perform FFT-type computations. With the advent of submicron VLSI technology, the smaller device geometries will afford either the implementation of greater functionality and/or computational capabilities. These increased capabilities will enable the solution of more demanding signal processing problems where there exist broad dynamic range requirements or low signal to noise conditions. But also associated with the increased functionality is the greater need for fault tolerance in high performance applications. Our bit-serial floating-point complex multiplier-accumulator will serve many such signal processing needs.

In recent years one and two dimensional systolic arrays have been designed to implement a wide range of matrix operations and signal processing algorithms. This paper proposes a three dimensional systolic array architecture of N3processors to further extend the performance advantages which can be achieved through regular local data transfer. The three dimensional array is discussed in the context of fixed point matrix-matrix multiplication which requires O(N3) multiplications and additions. The array pipelines N of these problems to attain a throughput rate which is practically independent of N. The performance, size, and fault tolerance of the array are discussed for the case N=32.

The Cornell University Signal Processor (CUSP) is a high performance CMOS processor which has been custom designed in VLSI to efficiently compute digital signal processing algorithms based on the Cooley-Tukey Radix-4 Fast Fourier Transform. One CUSP chip can be used as a stand-alone peripheral in a microprocessor system, or CUSP units can be combined into arrays in order to process signals with sampling rates of several MHz. It can attain a high numerical accuracy while maintaining a throughput superior to most available systems. In this paper we discuss the signal processing capabilities of CUSP, its simple programmability, and I/O interface. We also present simulation results which quantify the signal-to-noise ratio advantages of the bit-serial architecture.

This paper presents results on the analysis and digital hardware realization of a non-Gaussian impulse noise source which can be used to model the impulsive noise present in wide-band communications systems. The impulse noise appearing in practical systems is characterized by bursts of much higher amplitudes than would be predicted by a normal- or Gaussian-distribution law. A need exists for a better characterization of this non-Gaussian impulsive noise as well as actual noise generators which can be used for evaluation of present and future communications equipment under realistic system-operating conditions. The impulse noise model and the realization of the noise generator described in this paper are capable of producing Gaussian background noise, lognormal impulsive noise, and burst timing to set prescribed starting and ending times of the bursts and impulse density within each of the burst periods. The techniques used in this paper are based on first generating a basic pseudorandom binary sequence (PRBS) and then from this deriving the Gaussian background noise by adding samples from the PRBS generator. The lognormal noise is then derived from the same sequence by using an incremental exponential generator. The burst timing signals which realize an approximate Poisson distribution are obtained by observing the occurrences of specific subsequences from the PRBS generator.