Search

Proposes an optimization algorithm in which, instead of being minimized directly, the objective function in converted into a quadratic function of the filter coefficients whose minimum point can be obtained analytically. Two advantages of the cosine-modulated quadrature mirror-image filter (QMF); Information about conventional cosine-modulated QMF bank; Analysis on low-delay cosine-modulated QMF bank; Performance criteria of the design examples; Conclusions.

Proposes an algorithm for the design of the tree-structured vector quantizer that operates under storage and entropy constraints which is called the storage- and entropy-constrained tree-structured vector quantizer (SECTSVQ). Information about the SECTSVQ algorithm; Some experimental results on the application of the SECTSVQ to various sources; Conclusions.

Analyzes the convergence properties of the constrained and the unconstrained frequency-domain block LMS algorithms and gives theoretical analysis and computer simulation based on the comparison of both properties. Frequency-domain block LMS algorithms; Analysis of frequency-domain block LMS algorithm; Results of computer simulation to verify the analization; Conclusions.

The article discusses the simplified Newton-type adaptive estimation algorithms. Theoretical analysis of the proposed algorithm; Comparison of the proposed algorithm with LMS and RLS; Conclusion.

Establishes robustness, optimality, and convergence properties of the widely used class of instantaneous-gradient adaptive algorithms. The steepest descent and LMS algorithms; Information on the Cauchy-Schwarz inequality for vectors in an Euclidean space; Minimax optimality of gradient recursions; Conclusion.

Discusses about the first step toward the development and the evaluation of a new class of adaptive signal processing algorithms based on higher than second-order statistics. Background materials on cumulants of linear processes; How the elements of cumulant matrix are estimated; Results of the study to derived fast transversal recursive instrumental variable and generalized least mean square; Investigation on the convergence conditions; More.

Presents parallel algorithms on barrel shifter computers for a broad class of 1-D and 2-D signal operators. Reduced barrel shifter computer network; Parallel implementation of the I-D algorithms; Conclusion.

Presents a detailed description of the fast subspace tracking algorithm and its rank adaptive variation. Spherically averaged decompositions; Convergence; Complexity of various subspace tracking algorithms; Separation of the weakest spectral component; Problem statement.

Proposes a procedure based on bicepstrum iterative reconstruction algorithm for the reconstruction of an input signal of a linear-phase or zero-phase linear time-invariant system from its observed output. Signal expression; Image reconstruction; Cepstral coefficients.

Proposes a satisficing search algorithms for adaptively selecting near best basis and near best frame decompositions in redundant tree-structured wavelet transforms. Discrete packet transforms; Information costs functions; Basis selection methods; Time frequency analysis.

Addresses the problem of identifying the class of all multichannel stable system transfer functions that extend the given partial impulse response matrix sequence. Multichannel Schur algorithm; Parametrization of stable system; Rational system identification; Rational approximation of nonrational systems.

Presents an asymptotic performance analysis of three subspace-based methods for direction of arrival (DOA) estimation. Summary of ESPRIT algorithms; Asymptotic performance analysis of ESPRIT; Simulation and numerical evaluation; Conclusion.

States that scaled tangent rotation (STAR)-based recursive least squares (RLS) algorithm is designed such that fine-grain pipelining can be accomplished with little hardware overhead. Background and notation of problem formulation of adaptive filtering algorithm or QRD-RLS; Examination of the properties of STAR-RLS algorithm; Numerical properties of STAR-RLS; Stability and dynamic range analysis; Simulation results; Conclusion.

Investigates the transient and steady-state behavior of the quantized least-mean-squares (LMS) algorithm for Gaussian inputs. One of the most popular for real-time high-speed adaptive filters; Two major conclusions of the study; Steady-state analysis; Conclusions.

Presents a proof for the consistency of the w-slice algorithm in the estimation of the impulse response of an autoregressive moving average (ARMA) linear system. ARMA modelling; Numerical examples; Simulation results; Conclusions.

Presents an algorithm which offers a unifying study of the various two-dimensional rational transformations. Two-dimensional transformations of one-dimensional and two-dimensional polynomials; Computational complexity; Conclusion.

Presents an analysis of the partitioned frequency domain block least mean square (PFBLMS) algorithm. Eigenvalue spread problem due to the overlapping of successive blocks; Implementation of an adaptive transversal filter; Consequence of implementing the PFBLMS algorithm with a smaller overlapping of partitions; Computational complexity and memory requirement.

Presents a novel algorithm for calculating the running maximum or minimum value of a 1-D sequence over a sliding data window. Storage of a pruned ordered list of data elements that have the potential to become maxima or minima across the data window at some future time instant; Advantages over competing algorithms.

Presents a study of the quantization effects in the finite precision least mean square (LMS) algorithm with power-of-two step sizes. Deterministic nonlinear recursions for the mean and second-moment matrix of the weight vector; Wiener weight for white Gaussian data models and small algorithm step size mu; Monte Carlo simulations during all phases of the adaption process.

Considers a normalized version of the least mean square (LMS) algorithm with a decreasing step size mean(n) and prove the almost sure convergence of the nonhomogeneous algorithm. Formulation of the problem; Assumptions and convergence proof; Simulation example; Conclusion.

Introduces a class of adaptive algorithms for source separation that implements an adaptive version of equivariant estimation and is henceforth called equivariant adaptive separation via independence (EASI). Source separation; Serial matrix updating; Serial updates for orthogonal contrasts; Family of equivariant adaptive algorithms; Performance analysis of the adaptive algorithms; Numerical experiments of EASI; Conclusions.

Presents an efficient implementation algorithm for the Koilpillai-Vaidyanathan pseudo quadrature mirror filter (KVPQMF) bank which is useful in audio compression schemes. Derivation of the filter banks' analysis; Synthesis bank; How the KVPQMF bank can be implemented; Efficiency; Conclusions.

Presents a split-radix algorithm that can flexibly compute the Discrete Hartley Transforms (DHT) of various sequence lengths. Computational complexity of small DHT; Comparison of DHT's; Conclusions.

Focuses on the problem of finding the best time-varying filter bank tree-structured representation for a signal. Single tree algorithm; Double tree algorithm; Time-frequency algorithm; Flexible time segmentations; Comparison of the different algorithms; Experimental results of the study; Conclusion of the study.

Introduces the concept of underdetermined recursive least-squares (URLS) adaptive filtering. General framework in which some of parameter estimation algorithm can be derived; URLS-Fast transversal filter (FTF) algorithm; Efficient implementation of the URLS algorithm; Problems that may arise in the implementation of URLS; Simulation examples; Conclusion of the study.

Studies the performance of the recursive least squares (RLS) algorithm in the presence of a general chirped signal and additive white noise. Expressions for the optimum Wiener-Hopf coefficients; Filter decomposition; Filter misadjustment; Process of the chirped AR1; Simulations of the analysis; Conclusion of the study.

Introduces a new source separation technique exploiting the time coherence of the source signals. Formulation of problem; A second-order identification approach; Asymptotic performance analysis; Performance evaluation of the second-order blind identification (SOBI) algorithm; Conclusion of the study.

Analyzes the statistical performance of track initiation methods. Track initiation algorithm; Analysis of track detection probability; Analysis of false track probability; Conclusion of the study.

Presents an analytical treatment of the Mean-Square Error(MSE) performance for the leaky LMS adaptive algorithm for Gaussian input data. MSE analysis of the leaky LMS algorithm; Simulation examples from the analysis of the leaky LMS algorithm; Conclusions.

Presents an algorithm for deconvolution of sparse spike trains. Derivation of the algorithm; Implementation of the algorithm; Inverse signal-to-noise ratio; Computer simulation of the algorithm; Parallelization and recursive processing; Conclusions.

Examines the discrete-finite case of the multiwindow Gabor-type scheme. Issues of interest in the case of the discrete scheme (algorithms, complexity); Analysis of the property of discrete-finite multiwindow Gabor-type scheme; Calculation of algorithms for expansion coefficients and for the reconstruction of the signal and complexity analysis of the algorithms; Cases of single-window schemes and nonrectangular sampling of the combined space.

Part II. Examines asymptotic convergence speed of adaptive infinite impulsive response (IIR) filters of various structures and algorithms, including the simple constant gain (SCG) and Newton type. Background information about IIR filters; Math symbols and acronyms used in the study; Effects of finite precision (FP) on simple constant gain (SCG)-type algorithms; Information about cascade lattice and Newton type algorithms; Conclusions.

Compares four algorithms for adaptive retrieval of slowly time-varying multiple cisoids in noise, the adaptive notch filter, multiple frequency tracker, adaptive estimation scheme and hyperstable adaptive line enhancer. Algorithms for solving the recovery of noisy sinusoidal signals; Description of the four algorithms; Analysis of the algorithms to multiple cisoids; Numerical results of the study; Conclusions.

Considers the registration problem, which is a prerequisite process of a data fusion system to accurately estimate and correct systematic errors. Registration formulation used by real time quality control (RTQC), least squares (LS) and generalized least squares (GLS) methods; Development of the exact maximum likelihood (EML) registration algorithm; Asymptotic performance of the EML estimates; Effectiveness of EML algorithm; Conclusions.

Proposes a cumulant-based blind signal estimation algorithm for estimating the channel matrix in an n-sensor m-source system. Proposed cumulant-based estimation algorithm; Discussion of the proposed algorithm; Simulation results; Conclusion.

Proposes an algorithm for the resolution of overlapping echoes based on the constrained matched filters (CMF) that maximize the signal-to-noise ratio for a desired signal. Signal resolution and constrained matched filter; Resolution of overlapping echoes using CMF.

Proposes an efficient process to filter noise via an optimal stack filter. Review of the stack filter; Occurrence generation algorithm; Graphic searching algorithm; Output signal generation; Experimental results executed on IBM PC 486 machines; Conclusion.

Presents the robust modifications of the adaptive array algorithms with high robustness against jammer motion. Data-dependent derivative constraints; Robust algorithms with derivative constraints; Simulations of the algorithms; Conclusion.

Proposes an optimization criterion referred to as the joint optimization with subspace constraints (JOSC) to unify two types of approaches. Attractive features of JOSC algorithms; Performance and limitations of the two classes of blind estimators; Classification of moment-based algorithms by the JOSC criterion; Class of blind channel estimators based on the JOSC; Simulation results.

Presents an algorithm for obtaining nonnegative, joint time-frequency distributions satisfying the univariate marginals for the development of modern techniques to process stationary signals. Original work of Loughlin et al. for obtaining positive time-frequency distributions; Algorithms for obtaining positive time-frequency distributions; Three examples using the algorithms; Conclusions.

Proposes a two implementations of the least mean square (LMS)/Newton algorithm for efficient realization of long adaptive filters. Preliminaries that pave the way for the introduction of the algorithms; Analysis of the proposed algorithm; Simulation results and discussions; Results on application of the proposed algorithms to the implementation of an acoustic echo canceler.

Investigates the problem of direct frequency estimation in an adaptive notch filter. Adaptive algorithm for direct frequency estimation; Performance of adaptive algorithm; Comparison of the algorithm with some existing algorithm in terms of convergence speed, stability, robustness, and computation efficiency; Three numerical examples and corresponding simulations of algorithm.

Presents a detailed derivation of the proposed algorithms for the computation of both discrete fourier transform (DFT) of real data and the inverse DFT of the transform of real data. Computational structures of the algorithms for real data; Derivation of the radix-2 1 x 2 PM decimation-in-time (DIT) algorithm for computing RDFT; Computational complexity of the proposed algorithm.

Presents a practical method for the design of FIR filters in the complex domain. Several desirable properties of the algorithm; Numerical examples to illustrate the performance of the method; Conclusion.

Presents a detailed comparison of two on-line recursive least squares algorithms. Comparison of the computational load of recursive least square (RLS) and FHRLS; Computational load and memory requirements of the FHRLS algorithm; Numerical properties of RLS and FHRLS; Summary and conclusions.

Proposes a model for nonstationary time series analysis. Model evaluation and algorithm; Two important consequences of introducing such a model; Determination of parameters of basic model; General algorithm to choose good models.

Presents a bit-level pipelined very large scale integration (VLSI) architecture for the running order algorithm. Two main approaches for the linear array architecture; Modified algorithm suited to perform on a bit-level pipeline operation; Dedicated architecture with block processing for modified algorithm; Conclusions.

Determines all past input sequences that would have given rise to the variables in question, given the computed variables in a fast least-predication algorithm. Problem structure and related interpolation problem; Facilitation of formulas showing explicitly the form of all valid past inputs.

Presents an algorithm which can compensate for the phase rotation components introduced by frequency offsets in both far echo and far signal caused by the analog carrier network. Analysis of the method in terms of mean-square error; Estimation of the echo response; Simulation results which confirms the analysis.

Studies the properties of information matrices of the difference operator or the so-called delta operator-based algorithm for adaptive signal processing. Conditioning of a transformed information matrix in the delta domain for ill-conditioned problems; Advantages of using delta operator algorithms for adaptive signal processing that have been developed; Numerical examples to support analyses.