.png){kind=icon}
.png){kind=icon}
Return
Performance Optimization Using Delay in Multirate Reconstruction Systems
.png){kind=icon}
PRE
AI
DOI：10.1109/TCSI.2025.3619156.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
We consider the problem of designing a reconstruction system for multirate observations using the Wiener filtering framework. We model the observations as the output of a non-uniform analysis filter bank system and design a synthesis filter bank with an aim to achieve an overall delay system. Here, the reconstruction error depends not only on the synthesis filter bank used but also on the overall delay that is aimed for the system. For a given synthesis filter length, the process of obtaining the optimal delay that minimizes the mean-square of the reconstruction error is computationally expensive. In this paper, we demonstrate how a blocking operation provides a low-complexity approach to obtain an optimal delay in the Wiener filtering framework. We also derive an upper bound on the optimal delay, thereby reducing the search space. When the reconstruction error becomes zero, we obtain a perfect reconstruction (PR) system. For certain filter banks, multiple delay values may result in PR. We analyze the impact of synthesis filter length on this range of delay values. Finally, we present experiments to validate the theoretical results and compare the performance of our method with existing approaches. Our method achieves average execution time reductions of 67.34% and 19.55% for a system with a downsampling factor of 3, and 91.55% and 62.92% for a downsampling factor of 7, compared to the existing methods.
Keywords:
Multirate reconstruction system
multirate filter bank
Wiener filtering
reconstruction delay
perfect reconstruction
Journal
I
IF:
0
Papers:
268
Citations:
0
Organization
View more organizations