- 學位論文
SE列舉法與合併排序法在K-Best解碼之硬體架構及實現上的差異
The Difference Between SE-Enumeration And Merged Sort On The Hardware Architecture And Realization of K-Best Decoding
摘要
近年來無線通訊的生活化，多天線輸入輸出的技術越來越受到關注，主要的因素在於人們對無線通訊的需求越來越大，MIMO系統可以給予更高的效能。本論文採用K-Best演算法當作檢測的方式正是因為K-Best演算法犧牲一些準確度換取更高的效能。而在K-Best演算法中，如何快速的選取及排序是讓K-Best 演算法增加效率的關鍵之一。本論文選擇常見的合併排序法以及SE列舉法在複雜度及硬體上做比較。設定的環境是改良過的實數模型，使用的調變為16-QAM，K為4，並採用18位元。硬體的部分使用Xilinz ISE 12.2進行Verilog 的撰寫，再以Matlab 程式驗證其正確性，模擬的板子型號為Virtex 4系列中的xc4vlx160-12ff1513進行電路的合成。最後合成的結果SE列舉法無論在產出率或是面積大小都比合併排序法好上許多。
並列摘要
Recently, due to the wide application of wireless communication, MIMO system technology has attracted more and more attention. The main reason is that the demand of wireless communication is rising, and MIMO system can support higher throughput. In this thesis I use K-Best algorithm as a detective method, because K-Best algorithm can achieve higher throughput at a cost of lower accuracy. How to search and sort fast is a key to increasing throughput in K-Best algorithm. I choose merge sort and SE Enumeration in my thesis and compare their complexity and hardware. The set environment is real model with 16-QAM modulation; K is 4 and 18 bits. I use Xilinx ISE 12.2 to write Verilog code of the hardware and verify it with Matlab. At last, the simulation shows that SE Enumeration is better than merge sort in both hardware area and throughput.