使用基因演算法優化LTE-Advanced系統之不連續頻段載波聚合資源配置

為了配合快速成長的資料需求，滿足目前ITU所制定的 4G IMT – Advanced system為3GPP組織所制定的LTE-Advanced system，在LTE-Advanced系統其中一項關鍵性技術為Carrier Aggregation(CA)，CA是能夠讓基地台同時使用數種頻帶，這些頻帶可以是不連續且不同頻寬的頻帶。對於下路鏈結所採用OFDMA系統的頻帶， Transmission Time Interval(TTI)的所有資源區塊，經常和使用者的通訊品質SINR很相近。常見的資源配置方法Proportional Fair(PF)，針對每塊資源區塊分配完，就立即更新使用者的平均資料速率，所以適合於現在TTI的使用者，很可能就拿不到許多現在所適合的資源區塊，而分配給平均速率比重較低的使用者，非瞬時資料速率比重較高的使用者，導致降低資源配置的傳輸資料量。針對這問題，我們希望利用基因演算法可同時搜尋數個解空間、不易掉入區域解的特性，經過每世代演化，選出一條適合的資源分配結果，然後在不連續頻段的載波聚合場景，分析台灣地區的LTE-A系統頻帶，各種常見的資源配置方法，提出一套能夠透過維持一定公平性且最大化系統吞吐量之資源分配的演算法。根據模擬分析的結果，我們的方法可以提升相較於PF約12%的資料速率公平性以及2Mbps的系統吞吐量，相當於每位細胞邊緣使用者提升約0.14Mbps的資料速率以及每位細胞中心使用者提升約0.05Mbps，所以比起PF更適合使用在不連續頻段載波聚合的場景。

關鍵字

LTE-Advanced System ；載波聚合；基因演算法；資源配置

並列摘要

Carrier aggregation (CA) is proposed to comply the throughput performance required by the LTE-Advanced (LTE-A) system designed by 3GPP as the solution to the 4G requirements defined by ITU. CA enables the eNodeB to simultaneously utilize multiple frequency bands which can be non-contiguous and with different bandwidths. For the downlink OFDMA bands used for CA, the scheduling blocks (SB) during Transmission Time Interval (TTI) best matches to the current channel quality of user at given time slot. The Proportional Fair (PF) algorithm is popularly used as the resource allocation scheme which updates the average data rate of user right after resource allocation happens on every time slot of one TTI. However, for the user at the given time slot, the PF will not assign identical SB which best matches the to the current channel quality of that user at next time slot. Instead, the PF will allocate average data rate to low-priority users and non-instant data rate to high-priority users, which lowers the system throughput. Therefore, in this thesis, we utilized Genetic Algorithm (GA) to find a global solution among all users such that to optimize the system throughput by iteratively evolving the system parameters. To analyze the CA performance within Taiwan’s 4G spectrum, we then applied the result of GA with consideration of non-contiguous CA bands. The simulation result shows that the proposed GA-based system throughput optimization scheme outperforms PF in terms of data rate fairness about 12% and system throughput about 2 Mbps respectively, such that each cell-edge user and cell-center user can respectively enjoy 0.14 Mbps and 0.05 Mbps in terms of throughput increment accordingly. Therefore, we concluded that it is much more suitable to utilize GA for system throughput optimization than the legacy PF while using non-contiguous bands for the LTE-A CA function.

並列關鍵字

LTE-Advanced System ； Carrier Aggregation ； Genetic Algorithm ； Resource Allocation

參考文獻

[8] 鄭又維, " LTE-Advanced系統中基於載波聚合與使用者分組之比例式公平排程演算法," 國立臺北科技大學電腦與通訊研究所,碩士論文,July 2012

[21]林秉延, "以基因演算法同步決定一維構造之各層速度與厚度", 國立臺灣大學地質科學系, 碩士論文, July 2008

[20]林牧鼎, "在 LTE-Advanced架構下利用載波聚合技術與服務品質為基礎的排程演算法", 國立中央大學通訊工程學系, 碩士論文, July 2011

[19] 謝欣宏, "台鐵司機員排班與輪班問題之研究-以基因演算法求解", 國立成功大學交通管理科學研究所, 碩士論文, June 2002

[5] Na Guan, Yiqing Zhou, Lin Tian, Gang Sun and Jinglin Shi, "QoS guaranteed resource block allocation algorithm for LTE systems," Wireless and Mobile Computing, Networking and Communications (WiMob), 2011 IEEE 7th International Conference , pp.307,312, 10-12 Oct. 2011

國際替代計量

使用基因演算法優化LTE-Advanced系統之不連續頻段載波聚合資源配置

全文下載

主題瀏覽