By Topic

Statistical Signal Processing, 2003 IEEE Workshop on

Date 28 Sept.-1 Oct. 2003

Filter Results

Displaying Results 1 - 25 of 178
  • Decorrelating receivers of differential space-time coded CDMA systems

    Page(s): 319 - 322
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1439 KB) |  | HTML iconHTML  

    Capitalizing on code-division multiple-access (CDMA) and amicable orthogonal design, we propose a space-time (ST) transmission scheme for multiple-antenna systems, which allows full-diversity noncoherent communications and is resistant to multiuser interference (MUI). We then derive a differential decorrelating receiver (DDR) for flat Rayleigh fading channels, which decouples not only the detection of different users but also the decoding of different data symbols. The basic ideas of our design are as follows: decorrelating receiver eliminates MUI, and converts multiuser environments to single ones; while channel estimates are not available at the receivers, differential ST coding exploits ST diversity to improve the anti-fading performance of CDMA systems. Computer simulation results are given to support our development. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Bayesian blind PARAFAC receivers for DS-CDMA systems

    Page(s): 323 - 326
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1401 KB) |  | HTML iconHTML  

    In this paper an original Bayesian approach for blind detection for code division multiple access (CDMA) systems in presence of spatial diversity at the receiver is developed. In the noiseless context, the blind detection/identification problem relies on the canonical decomposition (also referred as parallel factor analysis [Sidiropoulos, IEEE SP'00], PARAFAC). The author in [Bro,INCINC'96] proposes a suboptimal solution in least-squares sense. However, poor performances are obtained in presence of high noise level. The recently emerged Markov chain Monte Carlo (MCMC) signal processing method provides a novel paradigm for tackling this problem. Simulation results are presented to demonstrate the effectiveness of this method. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Subspace based vowel-consonant segmentation

    Page(s): 589 - 592
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1511 KB) |  | HTML iconHTML  

    In our (knowledge-based) synthesis system [G. L. Jayavardhana Rama et al., 2002], we use single instances of basic-units, which are polyphones such as CV, VC, VCV, VCCV and VCCCV, where C stands for consonant and V for vowel. These basic-units are recorded in an isolated manner from a speaker and not from continuous speech or carrier-words. Modification of the pitch, amplitude and duration of basic-units is required in our speech synthesis system [G. L. Jayavardhana Rama et al., 2002] to ensure that the overall characteristics of the concatenated units matches with the true characteristic of the target word or sentence. Duration modification is carried out on the vowel parts of the basic-unit leaving the consonant portion in the basic-unit intact. Thus, we need to segment these polyphones into consonant and vowel parts. When the consonant present in any basic-unit is a plosive or fricative, the energy based method is good enough to segment the vowel and consonant parts. However, this method fails when there is a co-articulation between the vowel and the consonant. We propose the use of oriented principal component analysis (OPCA) to segment the co-articulated units. The test feature vectors (LPC-cepstrum & mel-cepstrum) are projected on the consonant and vowel subspaces. Each of these subspaces are represented by generalized eigenvectors obtained by applying OPCA on the training feature vectors. Our approach successfully segments co-articulated basic-units. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Multipath diversity and channel estimation using time-varying chirps in CDMA systems with unknown CSI

    Page(s): 335 - 338
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1430 KB) |  | HTML iconHTML  

    In this paper, we propose the use of wideband chirp signals in modulation schemes that combine multiuser detection with multipath diversity in frequency selective fading channels with known channel state information (CSI). We achieve this by appropriately designing the chirp parameters. For channels with unknown CSI, we also show that good channel estimation can be obtained, even in fast varying channels, with reduced interference by transmitting pilot chirp signals simultaneously with the users information. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Analysis of multicomponent LMF signals using time-frequency and the gray-scale inverse Hough transform

    Page(s): 190 - 193
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1477 KB) |  | HTML iconHTML  

    In this paper, the time-frequency distribution of multicomponent linear-frequency modulation (LFM) signals, which is a function of two variables, has been treated as an "image". The gray-scale inverse Hough transform (GIHT), an image processing method, is applied for analyzing the "image". The time-frequency distribution (TFD) is computed from an optimally weighted average of multiple Hermite windowed spectrograms, which provides a better "image source" than the commonly used Wigner distribution and spectrogram for the later image analysis. By taking advantage of GIHT more information in the time-frequency can be utilized and more accurate results have been achieved. Several examples are provided to demonstrate and quantify the effectiveness and robustness of this method. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Scheduling multiple sensors using particle filters in target tracking

    Page(s): 549 - 552
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1408 KB) |  | HTML iconHTML  

    A critical component of a multi-sensor system is sensor scheduling to optimize system performance under constraints (e.g. power, bandwidth, and computation). In this paper, we apply particle filter sequential Monte Carlo methods to implement multiple sensor scheduling for target tracking. Under the constraint that only one sensor can be used at each time step, we select a sequence of sensor uses to minimize the predicted mean-square error in the target state estimate; the predicted mean-square error is approximated using the particle filter in conjunction with an extended Kaiman filter approximation. Using Monte Carlo simulations, we demonstrate the improved performance of our scheduling approach over the non-scheduling case. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Iterative receivers for OFDM coded broadband MIMO fading channels

    Page(s): 355 - 358
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1431 KB) |  | HTML iconHTML  

    In this paper, we focus on the design of iterative receivers for OFDM coded systems in unknown broadband MIMO fading channels. The MIMO fading model is described by the environmental parameters such as mean angle of arrival, angular spread, antenna spacing, multiple clustered paths and Doppler shift. First, an expectation-maximization (EM)-based iterative receiver is presented for OFDM coded MIMO system without outer channel coding. The EM procedure is initialized by exploiting the time-domain correlation of the MIMO channels. When there is a mismatch on carrier frequency between the transmitter and receiver ends, which degrades the receiver performance, we also provide a scheme for frequency offset correction in MIMO systems. We further investigate a turbo-based EM receiver for OFDM coded MIMO system by adding interleaving and deinterleaving operators and outer channel coding. The receiver performance is proved to be greatly improved. For both systems, with and without outer channel coding, the performance degradation due to the spatial-domain correlation and/or frequency offset is presented in simulations. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A test for "long-memory" processes

    Page(s): 561 - 564
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1560 KB) |  | HTML iconHTML  

    This paper describes a multitaper test for long-memory processes against nonstationary alternatives. The test statistic is the largest canonical correlation between the multitaper eigencoefficients computed from data blocks offset in time. It is shown that the "Nile data" is not a long-memory process in the conventional sense, but has some persistent features at higher frequencies. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Root-exchange property of constrained linear predictive models

    Page(s): 90 - 93
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1420 KB) |  | HTML iconHTML  

    In recent works, we have studied linear predictive models constrained by time-domain filters. In the present study, studied the one-dimensional case in more detail. Firstly, we obtain root-exchange properties between the roots of an all-pole model and corresponding constraints. Secondly, using the root-exchange property we can construct a novel matrix decomposition ATRA# = I, where R is a real positive definite symmetric Toeplitz matrix, superscript # signifies reversal of rows and I is the identity matrix. In addition, there exists also an inverse matrix decomposition CTR-1C# = I, where C ∈ C is a Vandermonde matrix. Potential applications are discussed. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Characterization of wideband time-varying channels with multipath-scale diversity

    Page(s): 50 - 53
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1419 KB) |  | HTML iconHTML  

    In this paper, we present a characterization of wideband time-varying wireless channels based on a discrete time-scale framework that can provide multipath-scale diversity. This is important for systems such as underwater or airborne acoustics for which the narrowband conditions no longer hold, and time-scalings cannot be approximated by constant frequency shifts. We design a spreading waveform for direct-sequence code division multiple-access systems based on wavelet functions to realize orthogonal signaling. Furthermore, we use the orthogonal wavelet decomposition for direct channel estimation and coherent detection. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • New results on almost-sure identifiability of 2D-harmonic retrieval

    Page(s): 133 - 136
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1413 KB) |  | HTML iconHTML  

    In this paper the 2D harmonic retrieval problem is considered. New stochastic identifiability results are derived which equally holds true for the damped and undamped exponential mixtures. Previous results obtained for the 2D case indicate that up to K/2 L/2 exponentials can almost-surely be identified. In this contribution, we show that this bound is conservative. Simulations indicate that at least (KL)/3 harmonics can uniquely be resolved almost-surely. In the second part of the paper the obtained identifiability conditions are compared to stochastic uniqueness conditions of the 2D RARE algorithm. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A correlated signals MLE approach to joint source detection and direction of arrival estimation

    Page(s): 446 - 449
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1463 KB) |  | HTML iconHTML  

    The problem of detecting the number of (possibly correlated) narrowband sources of energy and estimating the direction of arrival (DOA) of each detected source using data received by an array of sensors is investigated. A combined detection and estimation approach based on the likelihood function (LF) is used. The approach is motivated by detection theoretic considerations instead of information theoretic criteria and uses maximum likelihood (ML) signal-to-noise ratio estimates of hypothesized sources as detection statistics rather than maximizing the likelihood function with a penalty function. An important feature that distinguishes this approach is the ability to trade off detection and false alarm performance, which is not possible with the other LF-based approaches, while achieving performance levels comparable to or exceeding the LF-based and MVDR approaches. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A bootstrap model selection criterion based on Kullback's symmetric divergence

    Page(s): 494 - 497
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1421 KB) |  | HTML iconHTML  

    Following in the recent work of J. Cavanaugh (1999) and A.K. Seghouane (2002), a new corrected variant of KIC develop for the purpose of sources separation is proposed in this paper. The variant utilizes bootstrapping to provide an estimate of the expected Kullback-Leibler symmetric divergence between the model generating the data and a fitted approximating model. Simulation results that illustrate the performance of the new proposed criterion for the detection of the number of signals received by a sensor array are presented. As a result, the KIC variant serves as an effective tool for estimating the number of sources compared to other well known criteria. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Renyi entropy based divergence measures for ICA

    Page(s): 565 - 568
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1419 KB) |  | HTML iconHTML  

    Information measures based on Renyi entropy provide a distance measure among a group of probability densities with tunable and flexible parameters to allow differentially granular differences in data. We interpret a recently developed measure, a α-JR divergence, as an alternative to mutual information (MI). We also present in this paper, its potential as an improved ICA criterion, and demonstrate its performance. We also propose a computationally efficient technique to approximate Renyi mutual divergence and apply it to analyze dependent data. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Blind adaptive multiuser detection for uplink long-code CDMA systems

    Page(s): 343 - 346
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1404 KB) |  | HTML iconHTML  

    In the uplink of a long-code CDMA system, base station knows spreading codes of all serviced users. Given propagation delays, a blind adaptive CMA-based approach, with the aid of a set of MMSE-like constraints parameterized by channel-like vectors, is proposed to detect all users' symbols simultaneously. As by-products, the channel-like vectors are also obtained for all users. Since the constraints involve covariance of the received data, which is time varying and can not be obtained by traditional sample average, we thus propose to approximate it using estimated signature matrix and noise power. Simulation results show satisfactory performance of the proposed method. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Bayesian learning for models of human speech perception

    Page(s): 408 - 411
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1479 KB) |  | HTML iconHTML  

    Human speech recognition error rates are 30 times lower than machine error rates. Psychophysical experiments have pinpointed a number of specific human behaviors that may contribute to accurate speech recognition, but previous attempts to incorporate such behaviors into automatic speech recognition have often failed because the resulting models could not be easily trained from data. This paper describes Bayesian learning methods for computational models for human speech perception. Specifically, the linked computational models proposed in this paper seek to imitate the following human behaviors: independence of distinctive feature errors, perceptual magnet effect, the vowel sequence illusion, sensitivity to energy onsets and offsets, and redundant use of asynchronous acoustic correlates. The proposed models differ from many previous computational psychological models in that the desired behavior is learned from data, using a constrained optimization algorithm (the EM algorithm), rather than being coded into the model as a series of fixed rules. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A multiresolution approach to pattern recognition

    Page(s): 417 - 420
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1453 KB) |  | HTML iconHTML  

    This paper reports on a family of computationally practical classifiers called dyadic classification trees (DCTs). Like many multiresolution methods in other application areas, DCTs are formed by recursive dyadic partitioning of the input space, followed by pruning to avoid overfitting. We investigate three pruning rules, each motivated by statistical learning theory. These pruning rules involve penalties that are non-additive, data-dependent, and scale-dependent. They produce learning rules that achieve near-minimax rates of convergence for a certain class of problems defined in terms of the smoothness of the Bayes decision boundary. Efficient algorithms exist for implementing each pruning rule. We then briefly mention an extension of dyadic classification trees that places polynomial decision boundaries at each leaf node. This polynomial decorated dyadic classification trees achieve faster rates for smoother decision boundaries, and have improved approximation capabilities relative to classifiers that employ a single polynomial decision rule, such as polynomial-kernel SVMs. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Adaptive nonlinear multigrid inversion with applications to Bayesian optical diffusion tomography

    Page(s): 170 - 173
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1450 KB) |  | HTML iconHTML  

    We previously proposed a general framework for nonlinear multi-grid inversion applicable to any inverse problem in which the forward model can be naturally represented at differing resolutions. The method has the potential for very large computational savings and robust convergence. In this paper, multigrid inversion is further extended to adaptively allocate computation to the scale at which the algorithm can best reduce the cost. We applied the proposed method to solve the problem of optical diffusion tomography in a Bayesian framework, and our simulation results indicate that the adaptive scheme can improve computational efficiency in this application. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Adaptive ML signal detection in non-Gaussian channels

    Page(s): 54 - 57
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1442 KB) |  | HTML iconHTML  

    The problem of robust signal detection in non-Gaussian noise is revisited. In this paper, we look at some issues of robust estimators which have been discussed very little in previous works. Some robust estimators, which are adaptive in nature and asymptotically efficient, are introduced and some technical improvements are suggested. Performance of these robust estimators is given in a practical communication problem and their asymptotic properties are investigated when the parameter-to-observation ratio becomes large. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Eigenstructure based blind channel order determination and kernel estimation for MIMO systems

    Page(s): 426 - 429
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1366 KB) |  | HTML iconHTML  

    The problem of blind identification of a linear system is widely noticed by many researchers in diverse fields, including speech processing, image reconstruction and data communication. Many existing blind channel identification algorithms require accurate estimates of the channel order. This problem has been studied for single input single output (SISO) as well as single input multiple output (SIMO) systems and several algorithms have been developed. Linear multiple input multiple output (MIMO) systems are considered in this paper. A novel algorithm based on subspace projections is proposed for the determination of all different orders, the number of subsystems and the kernels up to an ambiguity of post multiplication with an invertible matrix. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 'Distributed' signal processing: new opportunities and challenges

    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1258 KB)  

    Summary form only given. We are at a novel crossroads in technology where we are witnessing the confluence of computing, communicating and networking. A number of exciting applications are both driving and being driven by this confluence, including low-power sensor networks, large-scale ad hoc wireless networks, and wireless multimedia transmission. Many of these applications demand a move away from classical centralized architectures and algorithms towards more decentralized and distributed ones. Signal processing plays a key role in this revolution- not in isolation but rather as a pivotal interdisciplinary systems component, intimately integrated with communications, information theory, coding theory, and networking protocols. Sensor networks represent a particularly rich applications base. We would provide a snapshot of the sensor network related activities in a number of research groups at Berkeley. Motivated by the communications and computational constraints imposed by large-scale low-power sensor networks, we would describe some of our signal processing centric research including: (i) distributed sampling; (ii) distributed source coding; (iii) distributed estimation; and (iv) robust transmission. We would highlight the key foundational role played by multi-user information theory, particularly the so-called area of side-information coding for both source coding (compression) and channel coding (transmission). A deeper look reveals a beautiful functional duality between source and channel coding with side-information. This unexpectedly unifies a host of seemingly unrelated problem areas like distributed compression, digital watermarking, multimedia transmission over packet-error networks, and seamless digital upgrade of analog TV. Finally, as a microcosm of the expressive power of interdisciplinary thinking, we would describe a novel video compression paradigm dubbed PRISM (power-efficient, robust, hlh-compression, syndrome-based multimedia coding). PRISM's architecture, in stark contrast to that driving current video codecs like MPEG, allows for a novel shifting of the computational complexity from the encoder to the decoder, making it ideally suited for "uplink" transmission scenarios in wireless multimedia and surveillance a- pplications. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Sonar bottom mapping using fourth-order cumulants

    Page(s): 450 - 453
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1369 KB) |  | HTML iconHTML  

    The problem of sonar bottom mapping is addressed. Recent work has shown that high frequency seafloor backscatter can be characterized statistically by non-Gaussian, heavy tailed distributions. The statistical characterization can be exploited for improved bottom mapping. Known array processing methods are employed but with fourth-order cumulant functions rather than second-order covariance. The fourth-order statistics allow improved rejection of Gaussian interference for greater array gain in this application. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Image segmentation using factor graphs

    Page(s): 150 - 153
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1450 KB) |  | HTML iconHTML  

    Factor graphs were first studied in the context of error correction decoding and have since been shown to be a useful tool in a wide variety of applications. In this paper, we provide a brief introduction to factor graphs with an emphasis on their broad applicability, and then describe a new algorithm for segmenting binary images that have been blurred and corrupted by additive white Gaussian noise. Though the algorithm is developed for this particular class of images, generalizations are immediate. Simulation results detail the performance of the algorithm for images in three separate blurring conditions. The results suggest the potential for this approach, providing additional evidence of the usefulness of the factor graph framework. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Linear MMSE receiver using hidden training sequence in multipath channel for the W-CDMA system

    Page(s): 331 - 334
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1387 KB) |  | HTML iconHTML  

    A linear MMSE receiver using a hidden training sequence was proposed for the DS/CDMA system (S.Y. Jung and D.J. Park, 2003). The hidden training sequence, which uses a fraction of the informative sequence's transmitting power as a training sequence, was utilized for receiver design. In this paper, the proposed scheme is extended to the multipath channel case for the W-CDMA system. The performance of the proposed scheme is analyzed and confirmed through computer simulations. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Global sampling for sequential filtering over discrete state space

    Page(s): 498 - 501
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1438 KB) |  | HTML iconHTML  

    In many situations, it is required to approximate sequence of probability measures over a growing product of finite spaces. This is typically the case in digital communications, where the finite space is the symbol alphabet and the probability measures to be approximated are the posterior distribution of the transmitted symbols given the observations. Whereas it is in general possible to compute explicitly these probability measures, the typical complexity of these computations grow exponentially, precluding real time-implementations. In this paper, an efficient approach for approximating these distributions is presented using a particular implementation of the sequential Monte-Carlo filter (SMC). SMC consists in approximating the sequence of probability measures by the empirical distribution of a finite set N of trajectories which evolve under a random mechanism. Since the space is finite, it is possible to consider every offspring of the trajectory of particles: contrary to the classical sequential importance sampling and resampling (SISR) procedure, it is thus not required to develop a sophisticated strategy to build an appropriate importance distribution. The procedure is therefore straightforward to implement, and is well-suited for real-time implementation. The approach compares favorably with SMC techniques proposed in the literature and appears to be extremely robust even when the number of particles is small. An illustration on joint channel estimation / symbol detection on a flat fading channel is presented to support the claims. View full abstract»

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