| Title | Method for Estimating Amount of Delay of Sparse Channel Based on Phase-Only Correlation |
| Author | *Atsushi Imamizu, Kiyoshi Nishikawa, Hitoshi Kiya (Tokyo Metropolitan University, Japan) |
| Page | pp. 897 - 900 |
| Keyword | adaptive filter, POC, sparse system, LMS |
| Abstract | In this paper, we propose a method for estimating an amount of delay of a sparse system using phase-only correlation(POC). The performance of the normalized-least-mean-square(NLMS) algorithm is largely affected by sparse systems, namely its rate of convergence decreases. Although proportionate NLMS(PNLMS) was proposed to improve the performance when the unknown system is a sparse system, only the NLMS algorithm could be used to update the filter. Therefore its performance depends on the correlation of the input signal. By taking advantages of property of POC, we exploit the amount of delay of the unknown sparse system without the effect of the correlation. Simulation results show that the proposed method could provide better estimation of the amount of delay compared to PNLMS when the input signal is correlated, such as speech signals. |
| Title | Convex Combination of Affine Projection Filters with Individual Regularization |
| Author | *Kwang-Hoon Kim (Pohang University of Science and Technology, Republic of Korea), Young-Seok Choi (Johns Hopkins School of Medicine, United States), Seong-Eun Kim, Woo-Jin Song (Pohang University of Science and Technology, Republic of Korea) |
| Page | pp. 901 - 904 |
| Keyword | Affine projection algorithm, convex combination, regularization parameter |
| Abstract | We propose a convex combination of affine projection algorithm
(APA) filters with individual regularization. Two APA filters with
different regularization are adapted independently in order to
keep the advantages of both component filters. By using their own
error signals, separate filters are independently adapted, while
the combination is adjusted by means of a stochastic gradient rule
so as to minimize the error of the overall filter. This novel
scheme provides improvement of performance in term of the
convergence rate and the steady-state error. Experimental results
in channel identification show the validity of the proposed
algorithm when compared to the APA. |
| Title | Partial-Update Linfinity-Norm Adaptive Filtering Algorithm with Sparse Updates |
| Author | *Seong-Eun Kim (Pohang University of Science and Technology, Republic of Korea), Young-Seok Choi (Johns Hopkins School of Medicine, United States), Yun-Ki Han, Woo-Jin Song (Pohang University of Science and Technology, Republic of Korea) |
| Page | pp. 905 - 908 |
| Keyword | Adaptive filtering, Linfinity-norm, partial-update, sparse updates |
| Abstract | This paper provides a partial-update normalized
sign least-mean square (NSLMS) algorithm with sparse updates.
The proposed algorithm reduces the computational
complexity compared with the conventional L1-norm adaptive
filtering algorithms by decreasing the frequency of updating
the filter coefficients and updating only a part of the
filter coefficients. And we develop a mean square analysis to
present the convergence of the proposed algorithm. Experimental
results show that the proposed algorithm has the good
convergence performance with greatly reduced computational
complexity. |
| Title | A New Robust Narrowband Active Noise Control System Structure with Only One Channel for Referernce Signal Filtering |
| Author | *Yegui Xiao (Prefectural University of Hiroshima, Japan) |
| Page | pp. 909 - 912 |
| Keyword | Narrowband active noise control , Filtered-x LMS, Secondary-path, IIR notch filter, Computational cost |
| Abstract | In a conventional narrowband active noise control system, each
reference cosine wave has to be filtered by an estimate of the secondary-path.
We call this part {\it x-filtering} block.
The number of x-filtering blocks is equal to the number of targeted frequencies.
The computational cost of the system due to the x-filtering operations
may form a bottleneck in real system implementation.
In this paper, we propose a new narrowband ANC system structure which requires
only one x-filtering block regardless of $q$. All the cosine waves are
combined as an input to a x-filtering block whose output is decomposed by an
efficient bandpass filter bank into filtered cosine waves for the FXLMS that follows.
As a result, the computaional cost of the system is considerably reduced.
The new structure is further implanted in a recently developed ANC system
that is capable of mitigating the frequency mismatch (FM).
Simulations demonstrate that the new
systems present robust performance very similar to that of their counterparts,
but enjoy considerable advantages in system implementation. |
| Title | On-line and Off-line Based Approximation Algorithm for Model Predictive Control of Hybrid Systems |
| Author | Koichi Kobayashi, *Van Tang Nguyen, Kunihiko Hiraishi (Japan Advanced Institute of Science and Technology, Japan) |
| Page | pp. 913 - 916 |
| Keyword | model predictive control, hybrid systems |
| Abstract | In this paper, an approximate algorithm for model predictive control of hybrid systems is proposed. The proposed algorithm consists of the off-line and the on-line computations. In the off-line computation, lower and upper bounds of the optimal value of a given cost function are calculated. In the on-line computation, after the mode sequence is decided by using off-line computation results, the finite-time optimal control problem is solved. So the reduction of the computation time in the on-line computation is achieved. |