Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/8174
標題: 適用於多重標準具動態精確度調整之可伸縮性快速傅立葉轉處理器設計
Scalable FFT Processor Designs with Dynamic Precision Scaling for Multi-Standard Applications
作者: 曾俊堂
Tseng, Jiun-Tang
關鍵字: FFT;快速傅立葉轉換;Blocking Floating Point;OFDM;MIMO;區塊浮點數;正交分頻多工;多重輸入多重輸出
出版社: 電機工程學系所
引用: [1] Byung G. Jo, Myung H. Sunwoo, “New Continuous-Flow Mixed-Radix (CFMR) FFT Processor Using Novel In-Place Strategy,” IEEE Trans. On Circuits and Systems, Volume 52, Issue 5, pp.911 - 919, May 2005. [2] VLSI-oriented FFT algorithm and implementation,” ASIC Conference, pp.337 - 341, Sept. 1998. [3] Yu-Wei Lin, Hsuan-Yu Liu, Chen-Yi Lee, “A Dynamic Scaling FFT Processor for DVB-T Applications,” IEEE JOURNAL OF SOLID-STATE CIRCUITS, Volume 39, Issue 11, pp. 2005 - 2013, Nov. 2004. [4] Yun-Nan Chang, Keshab K. Parhi, “An Efficient Pipelined FFT Architecture,” IEEE Trans. On Circuits and Systems, Volume 50, Issue 6, pp.322 - 325, June 2003. [5] Haining Jiang, Hanwen Luo, Jifeng Tian and Wentao Song, “Design of an Efficient FFT Processor for OFDM Systems,” IEEE Trans. On Consumer Electronics, Volume 51, Issue 4, pp.1099 - 1103, Nov. 2005. [6] T. Sansaloni, A. Pe'rez-Pascual, V. Torres and J. Valls, “Efficient pipeline FFT processors for WLAN MIMO-OFDM systems,” Electronics Letters, Volume 41, Issue 19, pp.1043 - 1044, Sept. 2005. [7] Pandey, R., Bushnell, M.L., “Architecture for Variable-Length Combined FFT, DCT, and MWT Transform Hardware for a Multi-ModeWireless System,” International Conference on VLSI Design, pp.121 - 126, Jan. 2007. [8] Lee, H.-Y., Park, I.-C, “Balanced Binary-Tree Decomposition for Area-Efficient Pipelined FFT Processing,” IEEE Trans. on Circuits and Systems, Volume 54, Issue 4, pp.889 - 900, April 2007. [9] Guichang Zhong, Fan Xu, Willson, A.N., Jr., “A power-scalable reconfigurable FFT/IFFT IC based on a multi-processor ring,” IEEE Journal of Solid-State Circuits, Volume 41, Issue 2, pp.483 - 495, Feb. 2006. [10] Young-jin Moon, Young-il Kim, “A mixed-radix 4-2 butterfly with simple bit reversing for ordering the output sequences,” International Conference Advanced Communication Technology, Volume 3, pp.4, Feb. 2006. [11] Chua-Chin Wang, Jian-Ming Huang, Hsian-Chang Cheng, “A 2K/8K mode small-area FFT processor for OFDM demodulation of DVB-T receivers,” IEEE Trans. on Consumer Electronics, Volume 51, Issue 1, pp.28 - 32, Feb. 2005. [12] Sansaloni, T., Perez-Pascual, A., Valls, J., “Area-efficient FPGA-based FFT processor,” Electronics Letters, Volume 39, Issue 19, pp.1369 - 1370, Sept. 2003. [13] M. Hasan, T. Arslan, J.S. Thompson, “A Novel Coefficient Ordering based Low Power Pipelined Radix-4 FFT Processor for Wireless LAN Applications,” IEEE Trans. On Consumer Electronics, Volume 49, Issue 1, pp. 128 - 134, Feb. 2003. [14] Bernard, E., Krammer, J.G., Sauer, M., Schweizer, R., “A pipeline architecture for modified higher radix FFT,” International Conference on Acoustics, Speech, and Signal Processing, Volume 5, pp.617 - 620, March 1992. [15] Bidet, E., Castelain, D., Joanblanq, C., Senn, P., “A fast single-chip implementation of 8192 complex point FFT,” IEEE Journal of Solid-State Circuits, Volume 30, Issue 3, pp.300 - 305, March 1995. [16] Prabhu, J.A., Zyner, G.B., “167 MHz radix-8 divide and square root using overlapped radix-2 stages,” Symposium on Computer Arithmetic, pp.155 - 162, July 1995. [17] Yunho Jung, Hongil Yoon, Jaeseok Kim, “New efficient FFT algorithm and pipeline implementation results for OFDM/DMT applications,” IEEE Transactions on Consumer Electronics, Volume 49, Issue 1, pp.14 - 20, Feb. 2003.
摘要: 
針對以正交分頻多工為基礎之通訊系統,提出一種可伸縮性之快速傅利葉核心設計。此核心組成主要可分為兩種模組為Split-Radix-2/4/8 MDC模組與Radix2/4 MDC模組,使用兩個基本模組可以組成任意點數的FFT運算核心,加上回授的技術共用複數乘法器與旋轉因子查表電路減少面積的消耗,搭配前置資料穿插排程電路有效利用N/2點等待時間,並且對資料加以排程,同時使資料以穿插方式運算且改善傳統MDC架構在硬體使用效率上的不足,使整體核心之工作效率皆為100%。再加上區塊動態調整精確度的方式以群組的方式共同使用相同的指數,減少位移電路複雜度與只儲存指數的記憶體大小,經過模擬與實做結果證明可減少2位元的字元,減少10%晶片面積且能保持相同的量化誤差比,本論文中提出兩個架構一是最高支援512點scalable FFT核心以TSMC 0.18um 1P6M實現晶片面積為1.9x1.9mm2在100/50MHz執行速度下功率消耗為108.1mW,另一架構為最高支援2048點高傳輸率的scalable FFT核心,在102.8/51.4 MHz速度執行下其符合IEEE 802.15.3a標準要求409.6M Samples/sec的條件。

In this thesis, we propose a novel scalable FFT processor design capable of handling various transform sizes and data channels for multi-standard MIMO processing. The basic building blocks are split-radix 2/4/8 FFT modules in a modified MDC (multi-path delay commutator) form, pre-ordering data buffer, radix-2 butterfly module and twiddle factor module. Via proper data flow reconfiguration plus two alternative working frequencies, the processor can support various transform sizes plus 1 to 4 MIMO channels. The iterative compuaution scheme adopted in one of the split-radix 2/4/8 FFT module also facilitates the hardware sharing of complex multiplier and twiddle factor table. With the careful data scheduling in pre-ordering buffer, the hardware utilization of the FFT kernel is kept as 100% under various working modes. To reduce the circuit complexity, we further propose a block based dynamic scaling scheme. It can dynamically slide the position of data window by at most 2-bit. The width of line buffer and data path can thus be reduced accordingly. Despite the extra circuit employed in dynamic scaling scheme, simulation results indicate the overall chip area can be reduced by 10% without any additional quantization error loss. Finally, two versions of the scalable FFT processor designs using TSMC 0.18um 1P6M CMOS process are accomplished in this thesis. The first one can support from 64 up 512-point FFT computations The experimental results show that it has a core size 1.9x1.9mm2, and a power consumption of 108.1mW at 100/50MHz. The second design with parallel butterfly modules can support up to 2048-point computations and the highest processing speed can meet the throughput demand (409.6 MSamples/sec condition at 108.2/51.4MHz) IEEE 802.15.3a standard.
URI: http://hdl.handle.net/11455/8174
其他識別: U0005-1908200822461600
Appears in Collections:電機工程學系所

Show full item record
 

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.