Design and Implementation of Asymmetric Digital and Analog Filter Banks

M.Sc., L. Rosenbaum and Dr. H. Johansson

---

Background

This project is concerned with maximally decimated filter banks (FBs). Such FBs are frequently used in, e.g., digital communication, and speech, audio, and image processing applications. The main focus is on the design and implementation of asymmetric and near-PR FBs (see below). By design and implementation we mean that the FBs are designed not only to meet the specification at hand, but also in such a way that their arithmetic complexity, and thereby the power consumption, is minimized. Traditionally, the tendency has been that theory and design people take care of the first part whereas implementation people handle the second part. A goal in this project is to bridge this gap between different disciplines. In this way, the overall system performance can be improved.

Digital FBs have traditionally been designed in such a manner that the analysis FB and synthesis FB have the same arithmetic complexity. The objective function to minimize in these FBs is the overall complexity. However, in applications where it is important to minimize the complexity of either the analysis or the synthesis parts, the traditional FBs are not suitable. One such example is communication systems consisting of a stationary unit and a mobile unit where it is crucial to minimize the power consumption in the mobile unit whereas one can afford a somewhat higher power consumption in the stationary unit. Another example is mixed discrete-time and digital FBs for analog-to-digital conversion where the analysis filters are discrete-time filters such as SC-filters. In this case it is essential to minimize the complexity of the SC-filters since these are much more difficult to implement with a high accuracy compared to digital filters. We refer to FBs with different complexities in the analysis and synthesis parts as asymmetric FBs.

Further, the attention has traditionally to a large extent been paid to the problem of designing perfect reconstruction (PR) FBs for which the output is simply a shifted (delayed) version of the input. However, in practice FBs are most often used in applications where small errors (like quantization errors etc.) are allowed. Hence, it is rarely necessary to use PR FBs. Imposing PR on a FB is therefore an unnecessarily severe restriction which may lead to a higher arithmetic complexity than is actually required to meet the specification at hand. To reduce the complexity and power consumption one should therefore instead consider so called near-PR FBs which approximate PR.

Research

The main goals of this project are to gain a better insight into and understanding of the design and implementation issues of asymmetric filter banks and to develop new structures with low implementation cost. The goals can be summarized as follows.
 

• Insights into design and implementation issues of asymmetric and near-PR filter banks.
• New design techniques for low-power M-channel asymmetric and near-PR filter banks.
• New low-power implementation schemes for digital filter banks.

Results

Several different classes of cosine/sine modulated asymmetric and near-PR FBs as well as design techniques have been developed. These new FBs have lower complexity than previous FBs. Further, an advantage of the proposed FBs and design techniques, over many previously existing techniques, is that the FBs can easily be designed to meet general specifications.

Publications

[1] L. Svensson and H. Johansson, "Frequency-response masking FIR filters with short delay," in Proc. IEEE Int. Symp. Circuits Syst., Phoenix, USA, May 26-29, 2002.

[2] L. Svensson, P. Löwenborg, and H. Johansson, "Cosine modulated causal IIR NPMR and NPR filter banks," in Proc. Swedish System-on-Chip Conf., Falkenberg, Sweden, Mar. 18-19, 2002.

[3] L. Svensson and H. Johansson, "Narrow-band and wide-band frequency masking FIR filters with short delay," in Proc. National Conf. Radio Science (RVK), Stockholm, Sweden, June 10-13, 2002.

[4] L. Svensson, P. Löwenborg, and H. Johansson, "A class of cosine modulated causal IIR filter banks," in Proc. IEEE Int. Conf. Electronics Circuits Syst., Dubrovnik, Croatia, Sept. 15-18, 2002.

[5] L. Svensson, P. Löwenborg, and H. Johansson, "Asymmetric cosine modulated causal IIR/FIR NPR filter banks," in Proc. Second Int. Workshop Spectral Methods Multirate Signal Processing, Toulouse, France, Sept. 7-8, 2002.

[6] L. Rosenbaum, P. Löwenborg, and H. Johansson, "Modulated M-channel FIR filter banks utilizing the frequency response masking approach," in Proc. IEEE Nordic Signal Processing Symp., Hurtigruten, Norway, Oct. 4-7, 2002.

[7] L. Rosenbaum, P. Löwenborg, and H. Johansson, "Cosine and Sine modulated M-channel FIR filter banks utilizing the frequency response masking approach," in IEEE Int. Symp. Circuits Syst., Bangkok, Thailand, May 25-28, 2003.

Awards

L. Rosenbaum, P. Löwenborg, and H. Johansson received the best paper award at the 2002 IEEE Nordic Signal Processing Symposium with the paper "Modulated M-channel FIR filter banks utilizing the frequency response masking approach".

Members

• Linnea Rosenbaum, M.Sc., e-mail linnear@isy.liu.se.
• Håkan Johansson, Dr., e-mail hakanj@isy.liu.se, phone +46 13 281676.

 

Project funding

This project is funded by the INTELECT programme, which is financed by the Foundation for Strategic Research, SSF.

---

For information and questions regarding this web site, please contact Håkan Johansson, hakanj@isy.liu.se