適用於多輸入多輸出正交分頻多工系統之低複雜度部份層級映射內插式QR分解處理器

Reduced-Complexity Interpolation-based QR Decomposition Processor using Partial Layer Mapping for MIMO-OFDM Systems

柴立偉(Li-Wei Chai)

指導教授 ： 黃元豪

國立清華大學/電機資訊學院/通訊工程研究所/碩士(2011年)

摘要

近年來，在先進的多輸入多輸出正交分頻多工(MIMO-OFDM)系統中，對於速度的需求巨幅提升，而隨著FFT點數的提升，以及在多輸入多輸出系統中，更高的天線維度之需求，以往tone-by-tone的QR分解所產生的複雜度形成了實作上的瓶頸。為了解決此問題，內插式QR分解演算法已經被證實能有效的降低複雜度。在本論文中，我們提出一種低複雜度內插式QR分解演算法以減少在原始的內插式QR分解演算法的複雜度，除此之外，本論文也提出一種部分層級映射之技術。而我們我提出的低複雜度內插式QR分解演算法亦可和部分層級映射技術搭配，使得整體的複雜度能被更進一步的降低。同時，本論文也設計一種對應於低複雜度內插式QR分解演算法的硬體架構，並提出一種scaling scheme以解決在固定點數的情況下，原始的內插式QR分解演算法中會發生的動態範圍過大之問題。此外，本論文所提出的硬體架構具有適合平行處理之特性，可以適用於任何高傳輸速率的MIMO-OFDM系統中。最後，我們利用90nm UMC CMOS製程與Faraday cell library實現本論文所設計的低複雜度內插式QR分解處理器。透過完整的驗證流程後，在不考慮平行處理的條件下，本論文所設計的內插式QR分解處理器的速度可以達到45.6MQRD/s，而在平行處理的情況下，最高的處理速度則可以提升到182.4MQRD/s。

關鍵字

內插 ； 多輸入多輸出 ； QR分解

並列摘要

The throughput requirement of advanced MIMO-OFDM systems increases extremely in recent years. The complexity of traditional tone-by-tone QR decomposition raises along with FFT-point and MIMO dimension, and thus becomes the bottleneck of hardware implementation. The interpolation-based QR decomposition (IQRD) algorithm has been proved that it has lower complexity compared to traditional tone-by-tone QR decomposition algorithm.
In this thesis, a reduced-complexity interpolation-based QR decomposition (RC-IQRD) is proposed to decrease the complexity of original IQRD. Moreover, a low-complexity mapping scheme, called as partial layer-mapping (PLM) scheme, is adopted in RC-IQRD algorithm to further reduce the complexity. For the RC-IQRD algorithm, the corresponding architecture is proposed as well. Besides, we present a scaling scheme to solve the dynamic-range problem in IQRD algorithm. Nevertheless, the proposed architecture can use multiple hardwares easily to achieve higher throughput.
The proposed architecture is implemented by 90nm UMC CMOS technology and Faraday cell library. According to post-layout results, the proposed architecture can achieve 45.6MQRD/s with single QR decomposition unit, and the maximum throughput can be up to 182.4MQRD/s.

並列關鍵字

Interpolation ； MIMO ； QR decomposition

參考文獻

[1] E. Viterbo and J. Boutros, “A universal lattice code decoder for fading channels,” Information Theory, IEEE Transactions on, vol. 45, no. 5, pp. 1639 –1642, Jul. 1999.

[2] E. Agrell, T. Eriksson, A. Vardy, and K. Zeger, “Closest point search in lattices,”

[3] K. W. Wong, C. Y. Tsui, R. S. K. Cheng, and W. H. Mow, “A VLSI architecture of a K-best lattice decoding algorithm for MIMO channels,” in Proc. IEEE Int. Symp. On Circuits and Systems, vol. 3, May 2002, pp. 273–276.

[4] Y.-C. Liang, D. Kit, S. Attallah, W. S. Leon, and C. Xu, “Dynamic qrdm receivers

for mimo beamforming systems with imperfect channel state information,” 2005, pp. 801 –805.