基於隨機退後演算法輔助降低大規模機器傳輸時隨機存取碰撞之研究;Random Backoff Assisted Random Access Collision Reduction for Massive Machine-Type Cellular Communications

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/77801

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/77801

Title:	基於隨機退後演算法輔助降低大規模機器傳輸時隨機存取碰撞之研究;Random Backoff Assisted Random Access Collision Reduction for Massive Machine-Type Cellular Communications
Authors:	許哲維;Che-Wei, Hsu
Contributors:	資訊工程學系
Keywords:	大規模機器型通訊;先進長期演進技術;隨機存取;隨機後退;Massive Machine Type Communication(mMTC);Long Term Evolution(LTE);Random Access(RA);Random Backoff(RB)
Date:	2018-08-23
Issue Date:	2018-08-31 14:56:48 (UTC+8)
Publisher:	國立中央大學
Abstract:	技術與其他技術提供更多元網路服務是行動通訊技術所探討的議題。目前 LTE-A 通訊系統中，基站 (eNB) 僅對已完成 Radio Resource Control (RRC) 連線之使用者裝置 (UE) 進行分配上下行資源。當閒置狀態 UE 欲藉 eNB 上網時，須執行 Random Access Procedure (RAP) 建立 RRC 連線。此研究之動機來自觀察眾多 UEs 進行 RAP 時，一旦發生碰撞所造成之成本非常顯著。換言之，當 UE 於指定系統通道資源上傳送特殊前導訊號 (Preamble)， eNB 即可辨識 UE 之登入請求並分配上行資源給 UE 傳送 RRC 連線請求控制訊息。若多個 UEs 傳送相同 Preamble 則發生碰撞。由於 eNB 無法分辨是否有碰撞發生，這些 UEs 將同時於所配置之上行資源傳送個別 RRC 連線請求控制訊息，導致 eNB 無法解讀並透過 HARQ (Hybrid Automatic Repeat Request) 機制要求重送，後續重傳持續失敗直到 HARQ 重傳次數達到上限；之後，所有碰撞之 UEs 將重新執行 RAP 。而碰撞機率與 UE 數量呈正比；碰撞除了導致上線延遲外，亦造成無謂傳輸與耗電。5G 通訊系統已納入 massive MTC 之應用場景，如何降低大量 UE 進行 RAP 發生碰撞機率將是重要之技術。本研究以目前行動通訊網路規範進行探討，嘗試導入 WiFi 中 Random Backoff 演算法於 RAP 機制中，預期透過此設計降低 UE 間相互干擾。特別強調的是，此設計除了欲降低整體碰撞機率目標，同時考慮所提演算法與規範完全相容。 ;With the popularity of smart mobile devices, surfing the Internet anytime, anywhere has become a kind of necessity for modern life. How to combine the mobile communication technologies with other technologies to support more services in mobile networks is one of major issues. In current LTE-A network, the base station (eNB) only schedules the channel resource for the devices (UE) which has established the radio resource control (RRC) connections. For the UE staying in IDLE mode, it has to perform the random access procedure (RAP) in order to establish RRC connection with the eNB. The motivation of this thesis is based on the observations from the behavior and performance of RAP in machine type communications. When the number of UEs performing RAP excesses a certain threshold, collisions may occur and result in remarkable resource wastage and delay. In other words, a UE randomly selects one preamble and transmit it on the specified channel resource. The preamble is a specified sequence of a certain length and the system provides a number of orthogonal preambles for RAP. The eNB searches the preamble(s) and allocate uplink resource to UE(s) for each detected preamble. The corresponding UE then transmits RRC connection setup request message to the eNB on the uplink resource. If more than one UE sends the same preamble to eNB, their following messages will collide with each other and the eNB cannot decode any one of them. Then, the eNB utilizes the Hybrid Automatic Repeat Request (HARQ) feedback to notify UE(s) to retransmit message again. The collisions cannot be resolved and retransmissions will be lasted to the maximal HARQ retries. After then, all involved UEs restart RAP again. Obviously, the collision probability is linearly proportional to the number of UEs and more collisions will prolong the access delay and waste power and bandwidth. Due to 5G network has included the massive MTC (mMTC) usage case, how to efficiently reduce the collision probability in mMTC scenario could be an important and patentable technology. This project aims to deal with the contemporary mobile network system and try to integrate the RAP with the random backoff solution adopted in WiFi networks in order to minimize the interference among UEs. We also emphasize that the proposed scheme will not only reduce collision probability in RAP but also consider the full compatibility with current specifications.
Appears in Collections:	[Graduate Institute of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	290	View/Open

社群 sharing

Loading...