Pub Date : 2005-10-05DOI: 10.1109/TASSP.1987.1165054
H. Cox, R. Zeskind, Mark M. Owen
Adaptive beamforming algorithms can be extremely sensitive to slight errors in array characteristics. Errors which are uncorrelated from sensor to sensor pass through the beamformer like uncorrelated or spatially white noise. Hence, gain against white noise is a measure of robustness. A new algorithm is presented which includes a quadratic inequality constraint on the array gain against uncorrelated noise, while minimizing output power subject to multiple linear equality constraints. It is shown that a simple scaling of the projection of tentative weights, in the subspace orthogonal to the linear constraints, can be used to satisfy the quadratic inequality constraint. Moreover, this scaling is equivalent to a projection onto the quadratic constraint boundary so that the usual favorable properties of projection algorithms apply. This leads to a simple, effective, robust adaptive beamforming algorithm in which all constraints are satisfied exactly at each step and roundoff errors do not accumulate. The algorithm is then extended to the case of a more general quadratic constraint.
{"title":"Robust adaptive beamforming","authors":"H. Cox, R. Zeskind, Mark M. Owen","doi":"10.1109/TASSP.1987.1165054","DOIUrl":"https://doi.org/10.1109/TASSP.1987.1165054","url":null,"abstract":"Adaptive beamforming algorithms can be extremely sensitive to slight errors in array characteristics. Errors which are uncorrelated from sensor to sensor pass through the beamformer like uncorrelated or spatially white noise. Hence, gain against white noise is a measure of robustness. A new algorithm is presented which includes a quadratic inequality constraint on the array gain against uncorrelated noise, while minimizing output power subject to multiple linear equality constraints. It is shown that a simple scaling of the projection of tentative weights, in the subspace orthogonal to the linear constraints, can be used to satisfy the quadratic inequality constraint. Moreover, this scaling is equivalent to a projection onto the quadratic constraint boundary so that the usual favorable properties of projection algorithms apply. This leads to a simple, effective, robust adaptive beamforming algorithm in which all constraints are satisfied exactly at each step and roundoff errors do not accumulate. The algorithm is then extended to the case of a more general quadratic constraint.","PeriodicalId":282877,"journal":{"name":"IEEE Trans. Acoust. Speech Signal Process.","volume":"137 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2005-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133738508","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Task allocation and scheduling models for distributed digital signal processing are presented. The notions of block-type and stream-type tasks in signal processing application are introduced, and models for sequential and parallel I/O are presented. By extending the traditional models, more accurate schedules can be obtained. Those models can be further enhanced by allowing additional restrictions on the number of parallel I/O ports and the amount of parallelism on memory access. The deterministic nature of digital signal processing algorithms allows for more computationally intensive and accurate task allocation techniques to be performed at compile time. By applying a branch and bound algorithm, the task allocation problem can easily be solved for a variety of scheduling models and various system restrictions. >
{"title":"Task allocation and scheduling models for multiprocessor digital signal processing","authors":"K. Konstantinides, R. Kaneshiro, J. Tani","doi":"10.1109/29.61542","DOIUrl":"https://doi.org/10.1109/29.61542","url":null,"abstract":"Task allocation and scheduling models for distributed digital signal processing are presented. The notions of block-type and stream-type tasks in signal processing application are introduced, and models for sequential and parallel I/O are presented. By extending the traditional models, more accurate schedules can be obtained. Those models can be further enhanced by allowing additional restrictions on the number of parallel I/O ports and the amount of parallelism on memory access. The deterministic nature of digital signal processing algorithms allows for more computationally intensive and accurate task allocation techniques to be performed at compile time. By applying a branch and bound algorithm, the task allocation problem can easily be solved for a variety of scheduling models and various system restrictions. >","PeriodicalId":282877,"journal":{"name":"IEEE Trans. Acoust. Speech Signal Process.","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1990-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122719427","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The fast algorithm for the (real) Hartly transform is discussed in relation to the established fast algorithm for the (complex) Fourier transform. The two transforms are compared by timing comparably written programs on a given machine, and the discipline of timing is discussed as an adjunct to complexity analysis. With real data, one Hartley transform program can economically replace such packages as a complex-valued unilateral Fourier transform combined with a real-valued unilateral inverse Fourier transform. The Hartley transform is favorable for fast convolution of real data sets. The utility of spectral analysis into Fourier series throughout physics suggested that the Hartley transform might have less physical significance, but the construction of Hartley diffraction planes in the microwave and optical laboratories, where electromagnetic phase is encoded as real-valued field amplitudes, has revealed interesting complementarity. >
{"title":"Assessing the Hartley transform","authors":"R. Bracewell","doi":"10.1109/29.61544","DOIUrl":"https://doi.org/10.1109/29.61544","url":null,"abstract":"The fast algorithm for the (real) Hartly transform is discussed in relation to the established fast algorithm for the (complex) Fourier transform. The two transforms are compared by timing comparably written programs on a given machine, and the discipline of timing is discussed as an adjunct to complexity analysis. With real data, one Hartley transform program can economically replace such packages as a complex-valued unilateral Fourier transform combined with a real-valued unilateral inverse Fourier transform. The Hartley transform is favorable for fast convolution of real data sets. The utility of spectral analysis into Fourier series throughout physics suggested that the Hartley transform might have less physical significance, but the construction of Hartley diffraction planes in the microwave and optical laboratories, where electromagnetic phase is encoded as real-valued field amplitudes, has revealed interesting complementarity. >","PeriodicalId":282877,"journal":{"name":"IEEE Trans. Acoust. Speech Signal Process.","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1990-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134251570","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The total least squares (TLS) linear prediction (LP) method recently presented by Rahman and Yu (1987) and the equivalent improved Pisarenko's (IP) method by Kumaresan (1986) are reviewed and generalized by the whitening approach. The resulting whitened-TLS-LP method yields higher estimation accuracy than the TLS-LP. This simulation was carried out in double precision FORTRAN-77 on VAX-8810. The IMSL routines were used to perform eigendecompositions, compute the polynomial roots, and to generate the pseudo-Gaussian random numbers. >
{"title":"On the total least squares linear prediction method for frequency estimation","authors":"Y. Hua, T. Sarkar","doi":"10.1109/29.61547","DOIUrl":"https://doi.org/10.1109/29.61547","url":null,"abstract":"The total least squares (TLS) linear prediction (LP) method recently presented by Rahman and Yu (1987) and the equivalent improved Pisarenko's (IP) method by Kumaresan (1986) are reviewed and generalized by the whitening approach. The resulting whitened-TLS-LP method yields higher estimation accuracy than the TLS-LP. This simulation was carried out in double precision FORTRAN-77 on VAX-8810. The IMSL routines were used to perform eigendecompositions, compute the polynomial roots, and to generate the pseudo-Gaussian random numbers. >","PeriodicalId":282877,"journal":{"name":"IEEE Trans. Acoust. Speech Signal Process.","volume":"365 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1990-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133271644","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A system called net optimization and resource allocation (NORA) is introduced for the evaluation and programming of parallel signal processors, based on a data flow representation of the signal processing application. The main feature of this approach is that the scheduling and resource allocation can be done at compile time. It is made possible by the fact that most signal processing algorithms have constant data flow. The resulting hardware is much simpler because no overhead is needed for the real-time scheduling, as in usual data flow systems. Therefore a realization can easily be obtained using either commercially available components or VLSI technology. The proposed system comprises four main components: (1) a vector oriented data flow compiler for the translation of a high-level language description of algorithms into a data flow graph; (2) a critical path analysis for the evaluation of the minimal computation time of the algorithm, where block scheduling is assumed; (3) a schedule optimization for the determination of the minimal computation time under limited resources, not taking into account limitations imposed by the interconnection structure and temporary storage; and (4) a combined schedule optimization and resource allocation that maps a signal processing application onto a given hardware configuration and generates a formal microprogram. >
{"title":"A data flow technique for the efficient design of a class of parallel non-data flow signal processors","authors":"M. Thaler, G. Moschytz","doi":"10.1109/29.61543","DOIUrl":"https://doi.org/10.1109/29.61543","url":null,"abstract":"A system called net optimization and resource allocation (NORA) is introduced for the evaluation and programming of parallel signal processors, based on a data flow representation of the signal processing application. The main feature of this approach is that the scheduling and resource allocation can be done at compile time. It is made possible by the fact that most signal processing algorithms have constant data flow. The resulting hardware is much simpler because no overhead is needed for the real-time scheduling, as in usual data flow systems. Therefore a realization can easily be obtained using either commercially available components or VLSI technology. The proposed system comprises four main components: (1) a vector oriented data flow compiler for the translation of a high-level language description of algorithms into a data flow graph; (2) a critical path analysis for the evaluation of the minimal computation time of the algorithm, where block scheduling is assumed; (3) a schedule optimization for the determination of the minimal computation time under limited resources, not taking into account limitations imposed by the interconnection structure and temporary storage; and (4) a combined schedule optimization and resource allocation that maps a signal processing application onto a given hardware configuration and generates a formal microprogram. >","PeriodicalId":282877,"journal":{"name":"IEEE Trans. Acoust. Speech Signal Process.","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1990-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123793154","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The problem of determining the direction-of-arrival of narrowband plane waves using sensor arrays and the related problem of estimating the parameters of superimposed signals from noisy measurements are studied. A number of results have been recently presented by the authors on the statistical performance of the multiple signal characterization (MUSIC) and the maximum likelihood (ML) estimators for the above problems. This work extends those results in several directions. First, it establishes that in the class of weighted MUSIC estimators, the unweighted MUSIC achieves the best performance (i.e. the minimum variance of estimation errors), in large samples. Next, it derives the covariance matrix of the ML estimator and presents detailed analytic studies of the statistical efficiency of MUSIC and ML estimators. These studies include performance comparisons of MUSIC and MLE with each other, as well as with the ultimate performance corresponding to the Cramer-Rao bound. Finally, some numerical examples are given which provide a more quantitative study of performance for the problem of finding two directions with uniform linear sensor arrays. >
{"title":"MUSIC, maximum likelihood, and Cramer-Rao bound: further results and comparisons","authors":"P. Stoica, A. Nehorai","doi":"10.1109/29.61541","DOIUrl":"https://doi.org/10.1109/29.61541","url":null,"abstract":"The problem of determining the direction-of-arrival of narrowband plane waves using sensor arrays and the related problem of estimating the parameters of superimposed signals from noisy measurements are studied. A number of results have been recently presented by the authors on the statistical performance of the multiple signal characterization (MUSIC) and the maximum likelihood (ML) estimators for the above problems. This work extends those results in several directions. First, it establishes that in the class of weighted MUSIC estimators, the unweighted MUSIC achieves the best performance (i.e. the minimum variance of estimation errors), in large samples. Next, it derives the covariance matrix of the ML estimator and presents detailed analytic studies of the statistical efficiency of MUSIC and ML estimators. These studies include performance comparisons of MUSIC and MLE with each other, as well as with the ultimate performance corresponding to the Cramer-Rao bound. Finally, some numerical examples are given which provide a more quantitative study of performance for the problem of finding two directions with uniform linear sensor arrays. >","PeriodicalId":282877,"journal":{"name":"IEEE Trans. Acoust. Speech Signal Process.","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1990-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130579092","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Several modifications of the well-known LMS algorithm have been proposed for improved operation. This work analyzes one such algorithm that corresponds to the standard LMS algorithm with an additional update term, parameterized by the scalar factor alpha where mod alpha mod >
为了提高运算效率，对著名的LMS算法进行了一些改进。本文分析了一个这样的算法，该算法与标准LMS算法相对应，带有一个额外的更新项，参数化为标量因子alpha，其中mod alpha mod >
{"title":"Analysis of the momentum LMS algorithm","authors":"Sumit Roy, J. Shynk","doi":"10.1109/29.61535","DOIUrl":"https://doi.org/10.1109/29.61535","url":null,"abstract":"Several modifications of the well-known LMS algorithm have been proposed for improved operation. This work analyzes one such algorithm that corresponds to the standard LMS algorithm with an additional update term, parameterized by the scalar factor alpha where mod alpha mod >","PeriodicalId":282877,"journal":{"name":"IEEE Trans. Acoust. Speech Signal Process.","volume":"58 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1990-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134237420","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Analytical expressions for the Cramer-Rao bounds on the bearing and range of a Gaussian signal source observed by two-dimensional array in the presence of strong Gaussian interference are obtained. It is shown that: (1) all relevant features of array geometry are summarized by a function closely related to the conventional beam pattern; (2) the minimum signal-interference separation at which bearing estimation can be accomplished without serious loss of performance varies inversely with the first power of the signal-to-noise ratio; (3) in contrast to the localization problem in spatially incoherent noise, there is significant coupling between the estimation errors of bearing and signal power. Lack of prior knowledge of signal power can seriously degrade the quality of the bearing estimate; (4) the coupling of bearing and power estimates depends on the slope of the conventional beam pattern, not its magnitude. Control on sidelobe levels is therefore not sufficient to insure satisfactory localization performance; and (5) at large ranges (compared with the array dimensions) there is no coupling between the estimation of range and signal power. >
{"title":"Localization in the presence of coherent interference","authors":"H. Messer, Y. Rockah, P. Schultheiss","doi":"10.1109/29.61530","DOIUrl":"https://doi.org/10.1109/29.61530","url":null,"abstract":"Analytical expressions for the Cramer-Rao bounds on the bearing and range of a Gaussian signal source observed by two-dimensional array in the presence of strong Gaussian interference are obtained. It is shown that: (1) all relevant features of array geometry are summarized by a function closely related to the conventional beam pattern; (2) the minimum signal-interference separation at which bearing estimation can be accomplished without serious loss of performance varies inversely with the first power of the signal-to-noise ratio; (3) in contrast to the localization problem in spatially incoherent noise, there is significant coupling between the estimation errors of bearing and signal power. Lack of prior knowledge of signal power can seriously degrade the quality of the bearing estimate; (4) the coupling of bearing and power estimates depends on the slope of the conventional beam pattern, not its magnitude. Control on sidelobe levels is therefore not sufficient to insure satisfactory localization performance; and (5) at large ranges (compared with the array dimensions) there is no coupling between the estimation of range and signal power. >","PeriodicalId":282877,"journal":{"name":"IEEE Trans. Acoust. Speech Signal Process.","volume":"74 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1990-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132966248","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The problem of the parameter estimation of chirp signals is addressed. Several closely related estimators are proposed whose main characteristics are simplicity, accuracy, and ease of online or offline implementation. For moderately high signal-to-noise ratios they are unbiased and attain the Cramer-Rao bound. Monte Carlo simulations verify the expected performance of the estimators. It should be easy to extend this approach to signals having polynomials of any degree in the exponent. All the derivations will be done under the assumption that the signal-to-noise ratio is sufficiently high. >
{"title":"Parameter estimation of chirp signals","authors":"P. Djurić, S. Kay","doi":"10.1109/29.61538","DOIUrl":"https://doi.org/10.1109/29.61538","url":null,"abstract":"The problem of the parameter estimation of chirp signals is addressed. Several closely related estimators are proposed whose main characteristics are simplicity, accuracy, and ease of online or offline implementation. For moderately high signal-to-noise ratios they are unbiased and attain the Cramer-Rao bound. Monte Carlo simulations verify the expected performance of the estimators. It should be easy to extend this approach to signals having polynomials of any degree in the exponent. All the derivations will be done under the assumption that the signal-to-noise ratio is sufficiently high. >","PeriodicalId":282877,"journal":{"name":"IEEE Trans. Acoust. Speech Signal Process.","volume":"135 ","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1990-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120978069","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The acoustic-modeling problem in automatic speech recognition is examined with the goal of unifying discrete and continuous parameter approaches. To model a sequence of information-bearing acoustic feature vectors which has been extracted from the speech waveform via some appropriate front-end signal processing, a speech recognizer basically faces two alternatives: (1) assign a multivariate probability distribution directly to the stream of vectors, or (2) use a time-synchronous labeling acoustic processor to perform vector quantization on this stream, and assign a multinomial probability distribution to the output of the vector quantizer. With a few exceptions, these two methods have traditionally been given separate treatment. A class of very general hidden Markov models which can accommodate feature vector sequences lying either in a discrete or in a continuous space is considered; the new class allows one to represent the prototypes in an assumption-limited, yet convenient way, as tied mixtures of simple multivariate densities. Speech recognition experiments, reported for two (5000- and 20000-word vocabulary) office correspondence tasks, demonstrate some of the benefits associated with this technique. >
{"title":"Tied mixture continuous parameter modeling for speech recognition","authors":"J. Bellegarda, D. Nahamoo","doi":"10.1109/29.61531","DOIUrl":"https://doi.org/10.1109/29.61531","url":null,"abstract":"The acoustic-modeling problem in automatic speech recognition is examined with the goal of unifying discrete and continuous parameter approaches. To model a sequence of information-bearing acoustic feature vectors which has been extracted from the speech waveform via some appropriate front-end signal processing, a speech recognizer basically faces two alternatives: (1) assign a multivariate probability distribution directly to the stream of vectors, or (2) use a time-synchronous labeling acoustic processor to perform vector quantization on this stream, and assign a multinomial probability distribution to the output of the vector quantizer. With a few exceptions, these two methods have traditionally been given separate treatment. A class of very general hidden Markov models which can accommodate feature vector sequences lying either in a discrete or in a continuous space is considered; the new class allows one to represent the prototypes in an assumption-limited, yet convenient way, as tied mixtures of simple multivariate densities. Speech recognition experiments, reported for two (5000- and 20000-word vocabulary) office correspondence tasks, demonstrate some of the benefits associated with this technique. >","PeriodicalId":282877,"journal":{"name":"IEEE Trans. Acoust. Speech Signal Process.","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1990-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122651140","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：