在長期演進升級版網路上行多使用者多天線輸出輸入之傳輸排程演算法

Abstract

長期演進技術升級版(LTE-Advanced)為第三代合作夥伴計劃(3GPP)制定之標準，且被ITU被認定是4G標準。LTE-A和LTE一樣在下行鏈路中均使用正交分頻多工存取(OFDMA)技術，但在上行鏈路中使用LTE-A使用叢集式單載波分頻多工存取(clustered SC-FDMA)技術。相較於LTE的SC-FDMA，clustered SC-FDMA可以使系統排程更有彈性，因此LTE-A在上行傳輸部分可以比提供比LTE更快的傳輸速率。

為了使傳輸更快，我們採用了多使用者多輸出輸入系統(MU-MIMO)，MU-MIMO可以使一個資源區塊(RB)可以同時被多位使用者使用，因此可以增加系統的效能，在LTE-A MU-MIMO跟LTE MU-MIMO中最大的不同點是LTE-A採用了彈性UE配對(Flexible UE pairing)而LTE則是採用了固定UE配對(Fixed UE pairing)，採用彈性UE配對可以使在MU-MIMO中的UE配對更有彈性，也可以使系統效能更好。另外在LTE-A上行排程中，有兩個限制，第一個是使用者在配置資源時必須符合多段連續限制，第二個則是針對一個使用者在所有被配置的RBs都必須使用相同的調變技術(MCS)。此兩個限制使得LTE-A上行排程不能使用傳統的通道相依排程演算法(CDS)。 由於LTE/LTE-A上行鏈路的資源配置最佳解已經被証明為NP-hard問題，因此我們將藉由提出一個啟發式演算法，並且以模擬評估本論文提出之演算法，結果比其他篇論文的方法不論是效能還是公平性還來的佳。

Long Term Evolution Advanced (LTE-A) is a promising technology for 4G mobile networks standard by 3rd Generation Partnership Project (3GPP) and is regarded as 4G by ITU. LTE-A and LTE both utilize OFDMA in downlink, but LTE-A adopts cluster SC-FMDA in uplink. By comparison with SC-FDMA in LTE, cluster SC-FMDA make resource allocation more flexible, so LTE-A system can provides higher throughput than LTE system.

To increase the cell throughput, multi-user multiple-input multiple-output (MU-MIMO) are used. It can allocate multiple users in one Resource Blocks (RBs). The difference between LTE-A MU-MIMO and LTE MU-MIMO is that flexible UE pairing is adopt for LTE-A and fixed UE pairing for LTE。By using fixed UE pairing, the scheduler can pair UE more flexible to get better performance than fixed UE pairing. Channel Dependent Scheduling (CDS) may not perform well in LTE-A uplink due to its two inherent constraints－one is multi-contiguous RB assignment and the other is robust Modulation and Coding Scheme (MCS).

In this thesis, since the optimization problem of resource allocation in LTE/LTE-A uplink is NP-hard, we hence propose a heuristic algorithm, which will take not only the two inherent constraints but also MU-MIMO plus flexible UE pairing into consideration. We evaluate the proposed algorithm by conducting simulation. The simulation results show that our method can achieve more significant performance improvement in system throughput and fairness than other algorithms.