區塊空間調變(Block-based Spatial Modulation)，是一種在多輸入多輸出(Multi-Input Multi-Output)系統下的調變方式，此作法在每個時間點的傳送天線數量可以不同，可是在區塊裡，啟動天線數量總數要固定，本篇論文基於之前方法提出的模型做簡化，將區塊的大小定為2，也就是兩個時間點的啟動天線數總和要固定，調變的方式，將資訊流分成兩個部分，一部分決定啟動區塊的傳送向量，另一部分決定天線要傳送的信號，並將信號裝在區塊裡的啟動天線上，完成調變流程。
在這篇論文我們使用的解調方式為球狀解碼(Sphere Decoding)，接收端假設通道資訊(CSI)已知，有別於最大可能性偵測(Maximum Likelihood Detection)的高複雜度，做全搜尋的動作，球狀解碼的方式透過持續的降低搜尋半徑，大大的減少所要搜尋的範圍，同時錯誤率可以和ML解一樣好，本篇論文呈現球狀解碼使用在區塊空間調變的複雜度，以及球狀解碼使用在廣義空間調變(Generalized Spatial Modulation)的複雜度，將兩者各自的ML /SD複雜度比率做比較，可以發現將球狀解碼使用在區塊空間調變的複雜度降低比例大於廣義空間調變，所以比起將球狀解碼使用在廣義空間調變，將球狀解碼使用在區塊空間調變的效果更好。最後我們介紹一種名叫估計軟值(Estimated Soft)的方法，該方法可以產生一個半徑我們將他使用在球狀解碼的起始半徑上。與起始半徑無限制做出比較，我們可以發現其複雜度低於起始半徑無限制的球狀解碼。

Block-based spatial modulation (BSM) is a modulation method under the architecture of multi-input multi-output (MIMO) system. In this method, the number of transmit antennas at each time slot can be different, but the total number of active antennas in the block should be fixed. This thesis simplifies the model based on previous method and sets the size of the block to be two; that is, the total number of active antennas at two time slots must be fixed. In the method of transformation, the information flow is divided into two parts, one part determines the transmission vector of the active block, and the other part determines the signal to be transmitted by the antenna, and the signal is loaded into the active antenna in the block to complete the modulation process. In this thesis, we apply the sphere decoding (SD) as the demodulation method. The receiver assumes that the channel state information (CSI) is known. Compared with the exhaustive search to attain the maximum likelihood detection (ML) detection, the SD method reduces the search area by continuously reducing the search radius, and the error rate can be as good as the ML detection. We present the complexity of SD method used in block-based spatial modulation and the complexity of SD method used in the generalized spatial modulation (GSM). We also compare the ML/SD complexity ratios for these two methods. It can be found that the complexity reduction ratio of SD method in block-based spatial modulation is greater than generalized spatial modulation, so it is better to use the SD method for the block-based spatial modulation than for the generalized spatial modulation. Finally, we introduce a method called Estimated Soft, which can generate a radius and we use it as the starting radius of sphere decoding. Compared with an unlimited starting radius, we find that its complexity is lower than the SD algorithm with an unlimited initial starting radius.

[1] Shyam Gadhai, A. K. Sah, A. K. Singh, Rohit Budhiraja, A. K. Chaturvedi. “New Block-Based Spatial Modulation,”. IEEE Common. Lett, Vol:22, no. 10, pp.2016–2019, 2018.
[2] B. Hassibi and H. Vikalo. On the sphere-decoding algorithm I. expected complexity. IEEE Trans. Signal Process., Vol:53, pp.2806-2818, 2005.
[3] H.E1 Gamal M. O. Damen and G. Caire. On maximum-likelihood detection and the search for the closest lattice point. IEEE Trans. Inf, Theory, Vol:49, pp.2389-2402, 2003.
[4] Yang Jiang, Yeqiu Lan, Siyu He, Jinfu Li, Zeyu Jiang. Improved Low-Complexity Spere Decoding for Generalized Spatial Modulation. IEEE Commun. Lett, Vol:22, pp.1164-1167, 2018.
[5] Sara Jafarpoor, Majid Fouladian, Mohammad Neinavaie. Zero forcing based sphere decoder for generalized spatial modulation systems. ETRI, Vol:41, pp.145-159, 2019.
[6] João A. Cal-Braz, Raimundo Sampaio-Neto. Low-Complexity Sphere Decoding Detector for Generalized Spatial Modulation Systems. IEEE Commun. Lett, Vol:18, pp.949-952, 2014.
[7] Abdelhamid Younis, Sinan Sinanovic, Marco Di Renzo, Raed Mesleh, Harald Haas. Generalised Sphere Decoding for Spatial Modulation. IEEE Trans. On Communications, Vol:61, pp.2805-2815, 2013.
[8] Schnorr, C.P., Euchner, M. Lattice basis reduction. Improved practical algorithms and solving subset sum problems. Mathematical Programming 66, pp.181–199, 1994.
[9] 江承翰. 使用動態搜尋範圍來降低球狀解碼之複雜度. 國立成功大學電腦與通信工程碩士論文. 2018.