操作於在毫米波(Millimeter wave, mmWave)頻段的多輸入多輸出混合式波束合成系統可透過波束增益強化系統效能，但是系統的效能與波束合成產生的增益大小密切相關，所以波束的訓練以及波束搜尋對於建立良好的通訊連接就會顯得相當重要。在本篇論文中，系統所使用的通道環境是根據Millimeter-Wave Evolution for Backhaul and Access(MiWEBA)所建置的校園通道模型，並且我們採用階層的波束搜尋(hierarchical beam search)技術套用在寬頻的通訊系統中。首先，會先使用頂層的階層碼簿進行封包偵測(packet detection)，主要是透過頂層的階層碼簿其涵蓋的範圍可以達到全向性偵測的效果。當偵測到封包後，我們會切換至較低的階層碼簿並透過輸入的虛擬隨機(Pseudo-Random, PN)序列其良好的自相關特性(auto correlation)再來做進一步的波束搜尋與調整，最終在整體的效能以及搜尋的演算法中取得良好的平衡。我們所使用的在空間與時間的波束搜尋演算法，用於區分每個路徑的接收角度(Angle of arrival, AoA)、發送角(Angle of departure, AoD)以及路徑延遲。根據演算法完成硬體的實現，並透過硬體共享的方式來節省硬體的成本，最終以40奈米製程進行合成並操作於200 MHz的情況下，合成的面積為157856〖μm〗^2並總共花費5445個時脈下完成波束搜尋當中透過共享占整體面積較大的相關器以及能量計算器的部分可以幫我們省去31%的硬體資源，並且在透過使用階層式碼簿的情況下節省了50%以上的運作時間。;Millimeter wave (mmWave) multiple-input and multiple-output (MIMO) hybrid-beamforming systems are essential for recent wireless communications. However, the system performance relies heavily on beamforming gain. Thus beam training and beam search are critical to establish the communication link. In this paper, the campus channel model in WiWEBA is adopted. A hierarchical beam search algorithm is designed and implemented for wideband communication systems. In the beginning, multiple beams in the top level of a hierarchical codebook with coverage approaching omni-directional radiation pattern are used for packet detection. Then, PN-sequence is transmitted and the beam patterns in lower levels are utilized for further refinement to strike a good balance for performance and search complexity. A space-time beam search scheme is employed to distinguish the angle-of-arrival (AoA), angle-of-departure (AoD) and delay of each path. The corresponding hardware is also implemented. The hardware sharing technique is used to reduce the hardware cost. Finally, the hardware of space-time beam search is designed and verified in 40 nm CMOS technology. The area is about 157856 μm^2 while it takes about 5445 clock cycles for beam search at 200MHz clock frequency.
