深度学习辅助上行免调度NOMA多用户检测方法

针对上行免调度非正交多址接入（NOMA）场景中多用户检测的问题，通过结合传输数据的符号特征，提出基于深度神经网络（DNN）的联合活跃用户检测和数据检测框架. 考虑更一般化的实际场景，即用户在每个时隙中随机活跃. 将DNN求解结果作为改进的正交匹配追踪(OMP)算法先验输入，修正提升活跃用户检测和数据检测性能. 仿真结果表明，提出的多用户检测方案比传统的贪婪追踪及动态压缩感知（DCS）多用户检测算法具有更好的用户活跃性及数据检测性能.

Deep learning aided multi-user detection for up-link grant-free NOMA

A joint active user detection and data detection framework based on deep neural network (DNN) was proposed by combining the symbolic features of transmitted data in order to solve the problem of multi-user detection in uplink grant-free non-orthogonal multiple access (grant-free NOMA). The more general and practical scenario was considered, in which the user was randomly active in each time slot. The DNN solution result was used as a priori input of the modified orthogonal matching pursuit (OMP) algorithm in order to improve the user detection and date detection performance. The simulation results show that the proposed multi-user detection scheme has better user activity and data detection performance than the traditional greedy tracking algorithm and dynamic compressed sensing (DCS) multi-user detection algorithm.