LTE-A 異質網路下最小化延遲之傳輸點選擇

Abstract

異質性網路的支援在下一代通訊系統是一個相當重要的課題，其中的傳輸點選擇更是影響系統效能的關鍵因素。在本論文中，我們考慮在一個大型基地台的傳輸範圍內，有許多低功率遠端射頻收發裝置共同服務系統中的用戶。我們首先針對傳統的傳輸點選擇問題進行研究，在一個用戶只能被一個傳輸點服務的情況下，我們將問題抽象化成一個最佳化的問題。在此最佳化問題中，我們的目標函數是最小化所有用戶的傳輸延遲，而非一般常見的最大化系統容量。透過這樣的目標函數，除了可讓系統的總傳輸延遲明顯改善之外，還可以提高用戶之間的公平性。由於這樣的問題複雜度太高，於是我們提出了一套基於吉氏抽樣的快速演算法，並利用遠端射頻收發裝置可以和大型基地台即時交換資訊的特性，動態地為網路中的用戶找到合適的傳輸點，使效能趨近於最佳解。基於傳統傳輸點選擇問題的解法，我們進一步將多點協調傳輸與接收技術加入原先的問題中，允許有些用戶可選擇兩個以上的傳輸點來進行多點聯合傳輸。在我們的情境中，傳輸點彼此會形成不重疊的群組並且重複使用頻帶資源來為多點協調傳輸用戶服務。透過最佳化的分配，我們將系統的頻帶切割成兩個部份，其中多點協調傳輸用戶可在多點協調傳輸頻帶中選擇最佳的群集，而一般的用戶則在另一個頻帶中選擇最佳的傳輸點。從電腦模擬結果來看，在傳統的傳輸點選擇問題下，系統總傳輸延遲可以比原先減少約40%，公平性也較傳統的方法增進約50%；在加入多點協調傳輸後，系統總傳輸延遲可以更進一步降低約10%，公平性則可再有10%的提升。

Abstract Support of heterogeneous networks is a very important issue in next-generation LTE-Advanced systems. Transmission point selection in particular plays a key role in determining the performance of the system. In this thesis, we consider a scenario with a macro BS and several low-power RRHs within a cell. We start from the conventional transmission point selection problem, where each user can select only one transmission point for service. We formulate an optimization problem for transmission point selection by minimizing the total transmission delay. Compared with the conventional objective function for maximizing total system throughput, the formulated problem can significantly lower the total transmission delay while ensuring a certain degree of fairness among users. To solve the optimization problem, we propose a fast algorithm based on the concept of the Gibbs sampler and achieve a performance that is close to the optimal result. Built on the solution for the conventional transmission point selection problem, we then consider the scenario of coordinated multi-point transmission (CoMP), where users are allowed to select two or more transmission points for joint transmission and RRHs form non-overlapping clusters to reuse frequency resources. By dividing the available frequency band into the CoMP band and non-CoMP band, CoMP users and non-CoMP users can select individual transmission points for service. Simulation results show that, with the proposed transmission point selection algorithm, the total transmission delay can be reduced by 40% and fairness can be improved by 50% on average. The scenario with CoMP has an additional reduction of 10% in the total transmission delay and 10% improvement in fairness.