面向无授权传输的非正交多址接入技术研究

The next generation of wireless communication systems (5G) will focus on supporting internet of things (IoT) network. In order to meet the technical challenges of low latency, low signaling overhead and sparse packet arrivial, the academia and the industry propose to carry out uplink transmission using grant-free transmission: that is, when the packet arrives, no base station authorization is required, and on the contrary, random selection of multiple access mode is applied for uplink instant transmission. Unauthorized transmission results in multi-user interference, which naturally forms non-orthogonal multiple access. Existing NoMA technology can significantly improve spectrum efficiency under authorized transmission and increase the number of user access; however, there is no research on NoMA under no authorized transmission. This paper focuses on the NoMA core theory and key technologies for unauthorized transmission, including:(1) study the system level operation of grant-free transmission, as well as its application scenarios, (2) at bit level, study multi-level coding, which is suitable for grant-free transmission, (3)at symbol level, basing on the sparse spreading, the symbol mapping scheme of the joint code domain and the power domain is designed.(4) at receiver, based on the sparse characteristics of business as well as bit-level and symbol-level design, we focus on high-performance low-complexity receiver design.

未来物联网业务面临大规模接入、低延时、低信令开销和稀疏小包传输等技术挑战，无授权（grant free）的上行接入是应对该挑战的一种有效方法。使用无授权传输，用户侧无需基站授权即可发送数据。无授权传输导致多用户信号叠加，形成非正交多址接入（non-orthogonal multiple access, NoMA）。现有的NoMA能在授权传输时显著提高频谱效率、提升用户接入数量，但目前缺乏对无授权传输时NoMA的高效率接入方法和性能分析研究。本项目重点研究面向无授权传输的NoMA核心理论和关键技术，具体包括：（1）研究非正交无授权传输的高效接入方法及其性能分析；（2）研究适用于无授权传输的多层编码方法；（3）研究基于稀疏扩展的联合码域、功率域的符号扩展映射方法；（4）基于提出的高效接入方法，研究高性能低复杂度接收机。

伴随无线连接设备在各行各业中的爆发式增长，非正交多址接入（NOMA）已成为应对物联网等海量接入场景的核心手段。未来无线接入将面临更复杂多样的场景及性能需求，仅靠传统多址接入理论很难实现传输性能的进一步提升。基于无授权传输（grant free）的NOMA技术可使用户在数据包到达时随机挑选多址接入模式进行即时传输，无需基站授权，可有效应对多址接入中的低延时、低信令开销和稀疏小包传输等技术挑战。然而目前对无授权传输下NOMA的方法和性能研究还不充分，缺乏相应的低开销信令流程和高可靠收发机设计。为此，本项目基于变分优化和人工智能等方法，深入研究面向未来无线通信的无授权NOMA基础理论和关键技术，实现高可靠、低复杂度、低开销的多址接入。具体研究内容包括：（1）研究非正交无授权传输的接入、信令和传输模式配置，设计了一套无授权两步随机接入流程，降低了传输信令开销；（2）研究适用于无授权传输的叠加星座图，提出基于变分近似期望最大算法的星座旋转技术，逼近多用户信道容量界；（3）研究面向无授权多址接入的符号扩展映射模式，提出了包含随机用户激活、符号扩展的变分优化问题，基于人工智能技术设计了高效的多用户传输码本；（4）研究高性能低复杂度的多用户接收机，基于多用户信息论中的干扰删除结构提出了干扰删除神经网络（ICNN）用于多用户检测，降低了接收复杂度。