Báo cáo hóa học: Research Article Complexity-Aware Quantization and Lightweight VLSI Implementation of FIR Filters

Tuyển tập báo cáo các nghiên cứu khoa học quốc tế ngành hóa học dành cho các bạn yêu hóa học tham khảo đề tài: Research Article Complexity-Aware Quantization and Lightweight VLSI Implementation of FIR Filters
Nội dung trích xuất từ tài liệu:
Báo cáo hóa học: " Research Article Complexity-Aware Quantization and Lightweight VLSI Implementation of FIR Filters"Hindawi Publishing CorporationEURASIP Journal on Advances in Signal ProcessingVolume 2011, Article ID 357906, 14 pagesdoi:10.1155/2011/357906Research ArticleComplexity-Aware Quantization and LightweightVLSI Implementation of FIR Filters Yu-Ting Kuo,1 Tay-Jyi Lin,2 and Chih-Wei Liu1 1 Department of Electronics Engineering, National Chiao Tung University, Hsinchu 300, Taiwan 2 Department of Computer Science and Information Engineering, National Chung Cheng University, Chiayi 621, Taiwan Correspondence should be addressed to Tay-Jyi Lin, tjlin@cs.ccu.edu.tw Received 1 June 2010; Revised 28 October 2010; Accepted 4 January 2011 Academic Editor: David Novo Copyright © 2011 Yu-Ting Kuo et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The coefficient values and number representations of digital FIR filters have significant impacts on the complexity of their VLSI realizations and thus on the system cost and performance. So, making a good tradeoff between implementation costs and quantization errors is essential for designing optimal FIR filters. This paper presents our complexity-aware quantization framework of FIR filters, which allows the explicit tradeoffs between the hardware complexity and quantization error to facilitate FIR filter design exploration. A new common subexpression sharing method and systematic bit-serialization are also proposed for lightweight VLSI implementations. In our experiments, the proposed framework saves 49% ∼ 51% additions of the filters with 2’s complement coefficients and 10% ∼ 20% of those with conventional signed-digit representations for comparable quantization errors. Moreover, the bit-serialization can reduce 33% ∼ 35% silicon area for less timing-critical applications. coefficient quantization should take the filter complexity into1. Introduction consideration.Finite-impulse response (FIR) [1] filters are important In the literature, many works [19–29] have been pro- posed to obtain the discrete coefficient values such thatbuilding blocks of multimedia signal processing and wire-less communications for their advantages of linear phase the incurred additions are minimized. These works can beand stability. These applications usually have tight area classified into two categories. The first one [19–23] is to directly synthesize the discrete coefficients by formulatingand power constraints due to battery-life-time and cost the coefficient design as a mixed integer linear program-(especially for high-volume products). Hence, multiplier-less FIR implementations are desirable because the bulky ming (MILP) problem and often adopts the branch andmultipliers are replaced with shifters and adders. Various bound technique to find the optimal discrete values. Thetechniques have been proposed for reducing the number works in [19–23] obtain very good result; however, theyof additions (thus the complexity) through exploiting the require impractically long times for optimizing high-ordercomputation redundancy in filters. Voronenko and Püschel filters with wide wordlengths. Therefore, some researchers suggested to first design the optimum real-valued coefficients[2] have classified these techniques into four types: digit-based encoding (such as canonic-signed-digit, CSD [3]), and then quantize them with the consideration of filter com-common subexpression elimination (CSE) [4–10], graph- plexity [24–29]. We call these approaches the quantization-based approaches [2, 11–13], and hybrid algorithms [14, 15]. based methods. The results in [24–29] show that greatBesides, the differential coefficient method [16–18] is also amount of additions can be saved by exploiting the scalingwidely used for reducing the additions in FIR filters. These factor exploration and local search in the neighbor of thetechniques are effective for reducing FIR filters’ complexities real-valued coeffic ...