隨著通訊系統的發展，多輸入輸出系統(MIMO)[1]，逐漸變成技術的關鍵。這是由於MIMO 系統可以有效的改善系統的吞吐量(throughput)和傳送距離，並且不需要額外增加頻寬和發送功率，因此這項技術受到了廣泛的研究和運用。然而在解調MIMO 系統 的部分，如果使用一般常用的最大概似解調(MLD)會由於 MIMO 系統的特性，使得系統的複雜度將會隨著天線樹成指數成長。因此後續有了 ZF 和 MMSE這類能降低複雜度[2]的方法，但相對的會因此增加錯誤率。在本論文中，我們將會探討梯度搜尋如何在 MIMO 系統中的應用，首先我們會介紹何謂梯度搜尋理論[3]，接著討論初始序列對梯度搜尋的影響，接著利用球狀解碼求的初始序列後與 ZF、MMSE 進行比較，藉此說明這套理論跟一般梯度搜尋的差別。

With the development of communication systems, the Multiple Input Multiple Output (MIMO) system has gradually become the key technology. Since the MIMO systems can effectively improve system throughput, and does not need additional bandwidth and de-livered power, this technology has been extensively studied and used. However, for de-modulating the MIMO system, if the optimal maximum likelihood detection (MLD) is used, due to the characteristics of the MIMO system, the complexity of the system will grow exponentially with the number of antennas. Therefore, there are other sub-optimal methods such as zero forcing (ZF) and minimum mean-squares error (MMSE) detection that can reduce the complexity, but their error rates are also higher than that of the ML de-tection. In this thesis, we will study how the gradient search can be applied in MIMO sys-tems to improve the performance. First, we will introduce what is the gradient search algo-rithm, followed by the influences of the initial sequence on the gradient search. Finally, we use complexity-reduced sphere decoding to obtain the initial sequence, and we compare the performance when we apply the ZF and MMSE methods to obtain the initial se-quences. The simulation results show that when we apply the sub-optimal SD algorithm to obtain the initial sequence, the gradient search can obviously improve the performance.

[1] H. Bolcskei, "MIMO-OFDM wireless systems: basics, perspectives, and challenges," in IEEE Wireless Communications, vol. 13, no. 4, pp. 31-37, Aug. 2006.
[2] M. -X. Chang and W. -Y. Chang, "Maximum-Likelihood Detection for MIMO Sys-tems Based on Differential Metrics," in IEEE Transactions on Signal Processing, vol. 65, no. 14, pp. 3718-3732, 15 July15, 2017.
[3] M. -X. Chang and W. -Y. Chang, "Efficient maximum-likelihood detection for the MIMO system based on differential metrics," 2015 IEEE Wireless Communications and Networking Conference (WCNC), 2015.
[4] M. -X. Chang and W. -Y. Chang, "Efficient Detection for MIMO Systems Based on Gradient Search," in IEEE Transactions on Vehicular Technology, vol. 65, no. 12, pp. 10057-10063, Dec. 2016.
[5] F. Fisher and C. Windpassinger, “Real versus complex-valued equalization in V-BLAST systems,” IET Electron Lett., vol. 39, no. 5, pp. 470-471, Mar. 2003.
[6] S. Shafivulla, A. Patel and M. Z. A. Khan, "Low Complexity Signal Detection in MIMO Systems," 2018 IEEE 88th Vehicular Technology Conference (VTC-Fall), 2018.
[7] C. -Y. Hung and W. -H. Chung, "An improved MMSE-based MIMO detection using low-complexity constellation search," 2010 IEEE Globecom Workshops, 2010, pp. 746-750, 2010.
[8] B. Hassibi and H. Vikalo, "On the sphere-decoding algorithm I. Expected complexi-ty," in IEEE Transactions on Signal Processing, vol. 53, no. 8, pp. 2806-2818, Aug. 2005.
[9] B. Hassibi and H. Vikalo, "On the expected complexity of sphere decoding," Confer-ence Record of Thirty-Fifth Asilomar Conference on Signals, Systems and Computers (Cat.No.01CH37256), 2001.
[10] 吳沛樺. 改進球狀解碼降低複雜度之研究. Master's thesis, 國立成功大學﹐Jan2019.