應用於全雙工小細胞無線網路中的 最佳功率控制、波束成形與裝置選擇機制

Abstract

在本篇論文當中，在基地台是全雙工、使用者是半雙工的時候， 我們想要找出會讓整個系統的速率總和最大的波束成形向量。雖然自 我干擾消除技術在實際上的硬體限制是全雙工最大的挑戰，但是在最 近幾年，很多人提出了自我干擾消除技術在硬體設計上的提升，因此 全雙工的網路是可行的。接下來，我們假定經過抑制的自我干擾跟基 地台的傳送功率成正比的，然後我們把最佳化問題轉換成凸函式差異 的問題，根據弗蘭克-沃爾夫演算法的概念，在很短的次數，我們可 以漸進式得估計網路的效能，並且得到最好的波束成形向量跟最大的 速率總和。我們做了一些模擬，去計算全雙工網路的效能並且跟半雙 工的網路做比較。在學術與模擬結果中，可以發現如果自我干擾可以 有效消除的話，全雙工的網路效能會比半雙工的網路還要來得好。

In this thesis, we consider the problem to find the optimal beamforming vector that maximizes the sum rate of total system in which the base station works in full-duplex mode and the user equipments work in half-duplex mode. Although the key challenge for full-duplex is the practical hardware limitation in self-interference cancellation techniques, many improvements in hardware design for self-interference cancellation techniques have been reported in recent years. Therefore, the full-duplex network is feasible. Moreover, we assume self-interference after suppression is directly proportional to the transmission power of the base station. Then, we transform the optimization problem to the difference of convex (D.C.) programming problem. Based on the concept of Frank-Wolfe algorithm, we can iteratively approximate the throughput of the network and obtain the optimal beamforming vector and maximum sum rate in few iterations. We carry out some simulation and evaluate the performance of the full-duplex network compared with the conventional half-duplex network. Numerical results show that the fullduplex network outperforms the half-duplex network if the self-interference is substantially suppressed.