標題: | 高效率縮減搜尋範圍的平行處理之球型解碼器 An Efficient Admissible-set Elimination for Sphere Decoding with Parallel Processing |
作者: | 陳彥誠 Chen, Yan-Cheng 蔡尚澕 黃聖傑 Tsai, Shang-Ho Huang, Sheng-Chieh 電控工程研究所 |
關鍵字: | 球型解碼器;Sphere Decoding |
公開日期: | 2010 |
摘要: | 在這篇論文,我們提出一個縮小搜尋範圍以降低拜訪點數的方法。利用這方法,我們可以大幅的降低所需要拜訪的點數,因此也就可以更快找到解。除此之外,我們還提出一個不需要乘法器且平行處理的球型解碼器,因為平行處理,我們可以在把搜尋時間減少一半。因為我們是採用depth-first的搜尋方式,所以我們可以支援BPSK、QPSK、16-QAM和64-QAM各種不同的調變方式。因為我們有縮小搜尋範圍加上使用平行處理,所以即使是使用depth-first搜尋方式,也可以很快找到最佳解。在製程是TSMC 0.18um,SNR值20的時候,我們的吞吐量是105.7Mbps。 In this paper, we propose an admissible set elimination (ASE) algorithm to efficiently reduce the number of visiting nodes. With ASE, we can reduce the number of visiting nodes dramatically. In addition, we propose a multiplier-free sphere decoder with parallel processing units and use parallel processing units to greatly reduce the search time. Further, we can support BPSK, QPSK, 16-QAM and 64-QAM modulations, because each time we search only one path for every processing unit. With ASE and parallel processing units, we only searching a little nodes to find the maximum likelihood (ML) solution even if we search one path at one time. The maximum achievable throughput of our proposal can be up to 105.7Mbps for SNR=20 using TSMC 0.18um process. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#GT079712588 http://hdl.handle.net/11536/44482 |
Appears in Collections: | Thesis |