本文致力於具有混合式射頻/基頻預編碼設計之毫米波(millimeter-wave, mmWave)大型多重輸入多重輸出(massive multiple-input multiple-output, massive MIMO)通訊系統下的射頻電路(RF chain)對天線映射的方法,以最大化總和傳輸率(achievable sum-rate maximization, ASR maximization)為準則之最佳化預編碼設計。因在毫米波系統下,天線陣列數量非常大,以部分天線連接的方式,可以在傳輸效能與硬體成本之間取得一個平衡點。動態連接的方式可以使得最大總和傳輸率不會受制於固定的天線交錯式分配,並且趨近全連接架構的總和傳輸率。本文提出三個動態分配架構,使得射頻電路連接到較佳的天線子集,分別是曼哈頓距離法(Manhattan metric, MM)、組化相鄰相關法(contiguous subarray correlation, CSC)和等元素相鄰相關法(equally-neighboring correlation, ENC),經由電腦模擬得出,組化相鄰相關法有最低的迭代次數,達到同時降低複雜度、減少硬體使用功率又能維持一定總和傳輸率的目標。 This thesis concentrates on the radio frequency (RF) chain-antenna mapping operations in millimeter-wave (mmWave) massive multiple-input multiple-output (MIMO) systems with hybrid RF/baseband precoding/combining design. Three adaptive sub-array architectures offer a compromise between hardware complexity and spectral efficiency in massive MIMO networks at mmWave bands. These potentially dynamic RF chain-antenna planning methodologies comprise a variety of designs of the Manhattan metric (MM), contiguous subarray correlation (CSC), and equally-neighboring correlation (ENC). Remarkably, the CSC-based dynamically-connected configuration possesses excellent advantages of achievable sum-rate (ASR) performance and computational complexity simultaneously at a fixed hardware requirement in terms of phase shifters. Finally, simulation results demonstrate an adaptive partially-connected architecture is able to achieve an ASR capacity approaching the fully-connected design with an alleviated demand on both computation and hardware complexities.