Performance analysis of the DCT-LMS adaptive filtering algorithm

Kim, Dai I.; Wilde, Philippe De

doi:10.1016/s0165-1684(00)00098-0

Cited by 58 publications

(44 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…However, it is well known that, depending on the correlation level of the input data (i.e., eigenvalue spread of the input autocorrelation matrix), the LMS algorithm has its convergence rate compromised [1][2][3][4][5]. Aiming to overcome this drawback of the LMS algorithm, different strategies have been proposed and studied in the open literature [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19]. Such strategies generally lead to better performance at the expense of a substantial increase in the computational load (for details, see [5]).…”

Section: Introductionmentioning

confidence: 99%

“…[6,8,9,14,26,27]. On the other hand, the analysis of the TDLMS algorithm considering a nonstationary environment (i.e., a timevarying plant) has only been addressed in a few research works [15,25]. In [15], a modified power estimator is used in the TDLMS, leading to a model particularized for such a modified algorithm.…”

Section: Introductionmentioning

confidence: 99%

“…On the other hand, the analysis of the TDLMS algorithm considering a nonstationary environment (i.e., a timevarying plant) has only been addressed in a few research works [15,25]. In [15], a modified power estimator is used in the TDLMS, leading to a model particularized for such a modified algorithm. Moreover, [15] assumes that the output of each transformation branch is a sequence of statistically independent samples [16], which significantly compromises the accuracy of the obtained model (especially, in the transient phase).…”

Section: Introductionmentioning

confidence: 99%

“…In [15], a modified power estimator is used in the TDLMS, leading to a model particularized for such a modified algorithm. Moreover, [15] assumes that the output of each transformation branch is a sequence of statistically independent samples [16], which significantly compromises the accuracy of the obtained model (especially, in the transient phase). In [25], model expressions for predicting the steady-state excess mean-square error (EMSE) of some key algorithms (including the TDLMS) have been derived.…”

Section: Introductionmentioning

confidence: 99%

“…In the open literature, several approximations have been used to analytically determine (1), each of them having distinct accuracy levels [14,15,26,27]. In particular, better results are obtained in [27], where the Averaging Principle (AP) [28] is used to split (1) into two expected values, i.e.,…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Analysis of the TDLMS algorithm operating in a nonstationary environment

Kuhn

Kolodziej

Seara

2015

Digital Signal Processing

View full text Add to dashboard Cite

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Analysis of the TDLMS algorithm operating in a nonstationary environment

Kuhn

Kolodziej

Seara

2015

Digital Signal Processing

View full text Add to dashboard Cite

On the acceleration‐based adaptive inverse control of shaking tables

Dertimanis

Mouzakis

Psycharis

2014

Earthq Engng Struct Dyn

View full text Add to dashboard Cite

Accurate reproduction of time series with diverse frequency characteristics is a central issue in structural testing. This is true not only for simple experimental tests performed by reaction walls or shaking tables but also for more sophisticated ones, such as hybrid testing. Especially in the latter case, where actual feedback from an ongoing test is used in the calculation of the next excitation value, any possible mismatch may be fatal for both the validity of the test and the safety. The objective of this study is to propose a framework for the adaptive inverse control of shaking tables, which succeeds in this matching to a certain degree. By formulating a critical set of design specifications that correspond to safety, implementation, robustness and ease of use, the conducted research results in a design that is based on a modified version of the filtered-X algorithm with very competitive features. These are the following: (i) default operation in hard real-time and acceleration mode; (ii) very low hardware requirements; (iii) effective cancelation of the shaking table's dynamics; and (iv) robustness against specimen dynamics. For its practical evaluation, the method is applied to shaking table waveform replication tests under the installation of an approximately linear specimen of sufficiently high mass and complex geometry. The results are promising and suggest further research toward this field, especially in conjunction with hybrid testing, as the method retains certain global applicability attributes and it can be easily extended to other transfer systems, apart from shaking tables. A thorough review about actuator dynamics and related control issues (including the delay compensation problem) is given by Plummer [8]. Considering the transfer system in terms of systems theory, an ideal performance is depicted in Figure 1, in which the frequency response function (FRF) of a pure delay is illustrated. If the transfer system could be modeled by such a delay, then the input command (henceforth referred to as the reference signal) would be transmitted to the specimen unaltered (in the steady state), by introducing only an input-to-output delay. Such an assumption may be roughly valid for small-scale transfer systems, yet, this is not the case when complex, large-scale facilities of many mechanical parts are utilized for structural tests. This is particularly true for shaking tables, which require careful fine-tuning in order to take into account all the involved actuators (even when one-dimensional motion is applied), the corresponding kinematics, the resulted eccentricities of the table due to the installed specimen and the dynamics of the specimen itself. In their presentation of the controller for the E-Defense shaking table [9], Tagawa and Kajiwara show two indicative results about the tracking performance in both unloaded and loaded states, from where it follows that the controller approximates the 0-dB gain up to 2 Hz (three translational motions, unloaded case), reaching up to 4 Hz in the loaded case ...

show abstract

A New Variable Step Size Control Method for the Transform Domain LMS Adaptive Algorithm

Mayyas

2005

Circuits Syst Signal Process

View full text Add to dashboard Cite

This paper develops a new adaptive step size control method for the transform domain least mean-square (TDLMS) adaptive algorithm. The time-varying step size is an efficient approximation of an optimal step size based on a proposed cost function, therefore leading to significant improvements in the performance characteristics of the TDLMS algorithm. Mean-square analysis is provided, and an expression for the steadystate excess mean-squared error (MSE) is derived. Simulation experiments confirm the algorithm's enhanced convergence properties and verify the theoretical steady-state MSE results.

show abstract

Performance analysis of the DCT-LMS adaptive filtering algorithm

Cited by 58 publications

References 20 publications

Analysis of the TDLMS algorithm operating in a nonstationary environment

Analysis of the TDLMS algorithm operating in a nonstationary environment

On the acceleration‐based adaptive inverse control of shaking tables

A New Variable Step Size Control Method for the Transform Domain LMS Adaptive Algorithm

Contact Info

Product

Resources

About