隨著全球通訊技術的發展，第四代行動通訊系統(4G)已進入我們的生活之中。其中又以長期演進技術（Long Term Evolution, LTE）為代表。LTE使用了正交分頻多工(Orthogonal Frequency-Division Multiplexing, OFDM)以及多輸入多輸出(Multi-Input Multi-Output, MIMO)技術，使得第四代行動網路在上傳以及下載時，能擁有更大的傳輸量，及更遠的傳輸距離。 　　當下載時，因為每位使用者所在的環境不同，造成傳輸量、延遲時間因而不同，所以基地台分配資源時，尚有許多改善的空間。目前，許多文獻在探討如何公平且有效地分配資源塊(Resource Block, RB)給使用者裝置(User Equipment, UE)，如Proportional Fair (PF)與Modified Largest Weighted Delay First (MLWDF)兩種排程機制。前者考量了優先權的公平性，但沒有考量服務品質(Quality of Service, QoS)；後者對於較即時的封包有較高的優先權，卻導致資源分配不均，將大量的資源給部份使用者。本篇論文著重於公平性以及服務品質的考量，使用不同方式的佇列存放各種不同的資源型態(resource type)，依照資源型態佔整體資源數量的比例，以及計算出Adaptive Modified Largest Weighted Delay First (AMLWDF)的優先權值，針對各個UE與各個通道(channel)做全面性地配置，來提昇公平性及降低延遲以符合QoS要求。 / Owing to the development of global communication technology, the Long Term Evolution (LTE) is the latest technology for the fourth generation mobile communication systems (4G) that has entered into our lives. LTE uses Orthogonal Frequency Division Multiplexing (OFDM) and Multiple-Input Multiple-Output (MIMO) technology to provide high data rate transmission and long distance transmission when users doing download and upload. When doing download, users may have different throughput, delay time, and jitter due to they are in different locations. To improve these performance indexes, the E-UTRAN Node B (eNodeB) has to allocate resource blocks efficiently. In the literature, many works explore how to fairly allocate resource blocks (RB) to users. Proportional Fair (PF) and Modified Largest Weighted Delay First (MLWDF) are two example scheduling mechanisms. PF considers service priority and fairness, but doesn't consider the Quality of Service (QoS). MLWDF considers QoS but not service priority and fairness, and allows eNodeB giving more resources to particular users. In this thesis, we focus on resource allocation problem of downlink scheduling in LTE. Considering fairness and QoS, we store various resource types into particular queues and calculate the priorities using Adaptive Modified Largest Weighted Delay First (AMLWDF). The relationships between users and channels are coordinated according to the priorities for the sake of enhancing the fairness and reducing the delay time and jitter.

隨著無線通訊技術快速的發展，使用者對無線傳輸及品質的要求日益提升。第三代合作夥伴計劃3rd Generation Partnership Project (3GPP)所提出的長期演進技術Long Term Evolution (LTE)，是邁入第四代行動通訊系統(4G)的極佳選擇。其中以Single Carrier - Frequency Division Multiple Access (SC-FDMA)系統作為上行鏈路的主要通信技術。 然而，在上行SC-FDMA中有連續性資源塊配置的規定以及所配置之資源塊需使用相同調變技術的限制。一般而言資源分配的好壞會影響傳輸速率(data rate)與系統效能。因此，如何分配資源以改善傳輸速率與增進系統效能是本論文的研究重點。 目前，已有許多文獻在探討如何透過偵測頻率響應(frequency response)，解決資源分配的問題。但我們發現，透過偵測頻率響應的好壞進行優先權排班，無法有效符合現實網路服務的實際需求。在本論文中，我們擬以三階段的演算法來改善系統中資源分配的問題。第一階段，在time domain排班時根據不同的Quality of Service (QoS)特性區分User Equipment (UE)。在第二階段frequency domain排班時依據通道品質好壞決定優先權以分配無線電資源給UE。最後，我們根據SC-FDMA的兩大重要限制，提出在LTE上行系統中資源分配方法，在滿足Guaranteed Bit Rate (GBR)服務之時間延遲的限制時，亦能提升整體系統效能。 在模擬分析中，我們利用NS3進行模擬實驗分析，將我們所提出的三階段演算法在不同比例之網路服務類別的要求下，與固定子載波動態資源分配機制以及適應性動態子載波分配機制進行比較。實驗結果顯示，我們的方法在VoIP的average delay time相較於適應性動態分配法最多可改善約82.9%，real time gaming最多可改善約84.9%，而整體系統頻帶利用率(spectrum utilization)相較於固定與適應性分配機制最多可提升約15.3%。 / Long Term Evolution (LTE) is the latest standard of 3rd Generation Partnership Project (3GPP), which is one of the most promising technology for 4G mobile networks. The goal of LTE is to provide high data rate transmission, scalable bandwidth, low latency, and high-mobility. To achieve this goal, the LTE employs Orthogonal Frequency Division Multiplexing (OFDM) for downlink data transmission and Single Carrier - Frequency Division Multiple Access (SC-FDMA) for uplink transmission. This thesis focuses on the resource allocation problem of LTE SC-FDMA system. We propose a three-stage approach to improve resource allocation performance. In the first stage, we design a time domain scheduling according to different QoS features and time delay requirement to distinguish user equipment (UE). In the second stage, we design a frequency domain scheduling based on channel state information to give priorities to UEs. Finally, we propose resource allocation methods for LTE uplink under the two constraints of SC-FDMA. The proposed methods are proved to be able to meet the real-time service delay constraints and enhance overall system performance. In the simulations, the proposed three stage algorithms are compared to fixed sub-carrier dynamic resource allocation algorithm and adaptive dynamic sub-carrier algorithm against different proportions of network services. Simulation results show that our method outperforms the other two methods in terms of throughput, transmission delay and packet loss ratio.