5G上鏈之低延遲活動用戶識別與接收端處理

Abstract

為了實現大規模機器間之通信與實時的工業監控，低延遲一直是未來無線通信的關鍵設計問題。在本篇論文中，我們考慮以非正交多工技術來達到低延遲之兩通訊系統，壓縮式用戶識別系統為討論由多用戶對單一基地台傳輸所組成的上鏈傳輸系統，我們考慮在接收端疊加的訊號中找出活動用戶之身分，在頻率選擇通道之假設下，我們提出以壓縮感測為基礎的上鏈活動用戶偵測方案，達到降低延遲之效果，具體而言，每個用戶發送之前導信號並沒有貼尋環字首(Cyclic prefix)。在訊號還原方面我們使用21/-混合規範最小化(21/-mixed norm minimization)的凸優化問題(convex optimization problem)來還原活動用戶之身分與估計通道。透過區塊受限保距特性(block restricted isometry property)我們推導出更為嚴格的系統表現之上界。 另一個非正交多工技術稱為稀疏碼多工(Sparse Code Multiple Access，簡稱SCMA)是未來5G時代極具潛力的非正交多工技術，經由明智地設計傳送之稀疏碼字並在解調端使用信息傳遞演算法(Message Passing Algorithm，簡稱MPA)達到在高頻譜效率下依然有良好系統表現的特性。而SCMA技術之瓶頸主要在於MPA時現之複雜度。在本篇論文中，將介紹SCMA系統的運作原理，並詳細解釋MPA演算法的運作與實現方法以及提出一個基於門檻判定之MPA來降低複雜度；電腦模擬的結果將說明所提出方案之效能

To enable massive machine-type communication or real-time industrial monitoring and control for future 5G, low latency has been a key design issue for future wireless communication. In this thesis we consider two non-orthogonal multiple access systems for low-latency purpose. In the first system, finding the active user identities from the superimposed signals received at base station (BS) can be formulated. We proposes a CS-based uplink active user detection scheme over frequency selective channels, with the purpose of reducing latency. Specifically, during the user detection phase, the preamble sequence is transmitted directly without CP insertion. 21/-mixed norm minimization is used for user identification and channel estimation. A tighter upper bound of performance guarantee is derived by using block restricted isometry property (Block RIP). A commonly considered non-orthogonal multiple access is sparse code multiple access (SCMA), which has been identified as a potential solution for future 5G multiple access air interface. Through judicious design of sparse transmit codewords and by means of message-passing algorithm (MPA) implemented at the receiver, SCMA is capable of offering pretty good link quality at high spectral efficiency. A bottleneck for realizing SCMA into the current standards lies in the computationally demanding MPA. In the second part of this thesis, we will elaborate on how to implement an SCMA system and describe the process of MPA decoder in detail. A threshold-based technique is proposed to obtain a low-complexity implementation of MPA. Computer simulations are used to illustrate the performance of the proposed scheme.