2005IEEE Transactions on Signal ProcessingOpen access

LSE and MSE optimum partition-based FIR-IIR deconvolution filters with best delay

T. Engin Tuncer, Metin Aktaş

Open full text 12 citations

Abstract

We considered the least squares error (LSE) and mean square error (MSE) optimum best delay deconvolution filter design. We presented the partition-based finite impulse response-infinite impulse response (FIR-IIR) filters where the channel zeros are partitioned into two regions. The first region is composed of the selected channel zeros inside the unit circle, and the second region is composed of the remaining channel zeros outside the first region. Two methods for obtaining partitions are proposed, namely, optimum partitioning and ring-based partitioning. Resulting FIR-IIR filters are compared with the FIR and FIR-IIR unit circle best delay inverse filters in terms of their LSE. It is shown that the partition-based FIR-IIR filter performances are about 4-5 dB better than their FIR counterpart for the same complexity. Then, starting with the best delay FIR Wiener deconvolution filter formulation, we extended our results for the design of FIR-IIR MSE optimum deconvolution filters. We have compared the optimum-partitioning and ring-based FIR-IIR inverse filters with the FIR and IIR Wiener deconvolution filters. It is shown that FIR-IIR best delay inverse filters are less sensitive to the estimation errors compared with the IIR Wiener filters, and they perform better than the FIR Wiener filters. Furthermore, they are always causal and stable, making them suitable for real-time implementations.

About this research paper

What this paper is about

We considered the least squares error (LSE) and mean square error (MSE) optimum best delay deconvolution filter design. We presented the partition-based finite impulse response-infinite impulse response (FIR-IIR) filters where the channel zeros are partitioned into two regions. The first region is composed of the selected channel zeros inside the unit circle, and the second region is composed of the remaining channel zeros outside the first region. Two methods for obtaining partitions are proposed, namely, optimum partitioning and ring-based partitioning. Resulting FIR-IIR filters are compared with the FIR and FIR-IIR unit circle best delay inverse filters in terms of their LSE. It is shown that the partition-based FIR-IIR filter performances are about 4-5 dB better than their FIR counterpart for the same complexity. Then, starting with the best delay FIR Wiener deconvolution filter formulation, we extended our results for the design of FIR-IIR MSE optimum deconvolution filters. We have compared the optimum-partitioning and ring-based FIR-IIR inverse filters with the FIR and IIR Wiener deconvolution filters. It is shown that FIR-IIR best delay inverse filters are less sensitive to the estimation errors compared with the IIR Wiener filters, and they perform better than the FIR Wiener filters. Furthermore, they are always causal and stable, making them suitable for real-time implementations.

Why it matters

OpenAlex reports 12 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

We considered the least squares error (LSE) and mean square error (MSE) optimum best delay deconvolution filter design. We presented the partition-based finite impulse response-infinite impulse response (FIR-IIR) filters where the channel zeros are partitioned into two regions. The first region is composed of the selected channel zeros inside the unit circle, and the second region is composed of the remaining channel zeros outside the first region. Two methods for obtaining partitions are proposed, namely, optimum partitioning and ring-based partitioning. Resulting FIR-IIR filters are compared with the FIR and FIR-IIR unit circle best delay inverse filters in terms of their LSE. It is shown that the partition-based FIR-IIR filter performances are about 4-5 dB better than their FIR counterpart for the same complexity. Then, starting with the best delay FIR Wiener deconvolution filter formulation, we extended our results for the design of FIR-IIR MSE optimum deconvolution filters. We have compared the optimum-partitioning and ring-based FIR-IIR inverse filters with the FIR and IIR Wiener deconvolution filters. It is shown that FIR-IIR best delay inverse filters are less sensitive to the estimation errors compared with the IIR Wiener filters, and they perform better than the FIR Wiener filters. Furthermore, they are always causal and stable, making them suitable for real-time implementations.

Key concepts: Infinite impulse response, Finite impulse response, Mathematics, Wiener filter, Deconvolution, Inverse filter, 2D Filters, Wiener deconvolution

Related papers

Back to paper searchBrowse research topicsOriginal source
LSE and MSE optimum partition-based FIR-IIR deconvolution filters with best delay — Research Paper | ScholarLens