摘要: | 在此論文裡,我們提出了可適用於6×6、5×5、4×4、3×3 和2×2不同的天線組態,以及可支援64-QAM、16-QAM、QPSK和BPSK不同的調變方法且K值10的K最佳多輸入輸出解碼器。我們設計了改良型K最佳演算法(Modified K-best, 以下簡稱MKB)來取代傳統K最佳演算法每層都需要排序的特色,改為每兩層排序一次減少排序的次數,為了進一步減少拜訪的點數又設計了新式的編碼式列舉法(Code-Book Enumeration, 以下簡稱CBE)將拜訪節點由K×√M減少為K×e,針對排序效率的提升設計了新式的排序方法平行切割合併法(Parallel-Slice Merge Algorithm, 以下簡稱PSMA)和平行氣泡切割排序法(Parallel Bubble-Slice Sort, 以下簡稱PBSS)來加快排序的速度。在硬體的實現上,我們利用MKB與排序電路的組合方塊來達到管線式架構可配置的需求。為了減少乘法器的複雜度,我們使用移位乘法器(Shift Multiplier, SM)來取代傳統的乘法器。本論文使用SMIMS VeriEnterprise Xilinx FPGA驗證電路功能,並以Design Compiler 與UMC Faraday 90μm製程合成與其他文獻作比較。 In this thesis, we proposed a MIMO detector which can support multiple antenna types (6×6, 5×5, 4×4, 3×3, and 2×2), various modulation schemes (64-QAM, 16-QAM, QPSK, and BPSK) and K-value (K=10). From the algorithm aspects, we deigned a modified K-best algorithm (MKB) can reduce the number of sorting layer from 2Nt to Nt, compared with the conventional K-best algorithm. The MKB also include the Code-Book Enumeration to reduce number of visted nodes from K×√(M ) to K×e.To further enhance sorting performance, we design the Paralled-Slice Merge algorithm(PSMA) and Parallel Bubble-Slice sort(PBSS). In terms of hardware implementation, the MKB and sorting circuit are designed as elementary building blocks. With these building blocks, the proposed MIMO detector can flexibly achieve the configurable architecture. In order to simplify the multiplier complexity, we propose a novel shift-multiplier (SM) to replace the conventional multiplier. Finally, the proposed configurable MIMO detector is verified by the SIMIS VeriEnterprise Xilinx FPGA development board. |