Abstract: Internet of Things (IoT) communication system has the characteristics of small active user number and short data frame, while the pilot and user identification code required by channel estimation and multi-user detection will greatly reduce the communication efficiency and response speed of IoT system. To solve these problems, a blind channel estimation and multi-user detection algorithm based on Non-Orthogonal Multiple Access (NOMA) was proposed. Firstly, the spread spectrum matrix in Code Division Multiple Access (CDMA) system was used to allocate the carrier to each user, and the constellation rotation problem caused by blind estimation was solved by differential coding. Secondly, according to the sparsity of carriers allocated to users, the Bernoulli-Gaussian (B-G) distribution was introduced as a prior distribution, and the hidden Markov characteristic between the variables was used to perform the factor decomposition and modeling, and the multi-user identification was carried out based on sparse features of user data. Finally, the above model was deduced by message passing algorithm to solve multi-user interference caused by NOMA, and the joint channel estimation and detection receiver algorithm for IoT environment was obtained. The simulation results show that, compared with Block Sparse Single Measurement Vector (BS-SMV) algorithm and Block Sparse Adaptive Space Pursuit (BSASP) algorithm, the proposed algorithm can achieve a performance gain of about 1 dB without increasing the complexity.

Key words: Internet of Things (IoT) communication, blind channel estimation, probabilistic graph model, iterative receiver design, Non-Orthogonal Multiple Access (CDMA)

摘要： 物联网（IoT）通信系统具有活跃用户数低、数据帧短等特性，而信道估计和多用户识别所需的导频和用户识别码会显著降低IoT系统的通信效率和响应速度。针对上述问题，提出了一种基于非正交多址接入（NOMA）的盲信道估计和多用户检测算法。首先，利用码分多址（CDMA）系统中的扩频矩阵对每个用户所占用载波进行分配，并采用差分编码解决盲估计所引发的星座点旋转问题；其次，根据用户所分配载波具有的稀疏性，引入伯努利-高斯（B-G）分布作为先验分布，利用变量间的隐马尔可夫特性进行因式分解和建模，由用户数据的稀疏特征进行多用户识别；最后，应用消息传递算法对上述模型进行推导，解决NOMA引起的多用户干扰，得到面向IoT环境的联合信道估计和检测接收算法。仿真结果表明，相对于块稀疏单测量向量（BS-SMV）算法和块稀疏自适应子空间求解（BSASP）算法，所提算法能够在不提高复杂度的条件下获得约1 dB的性能增益。

关键词: 物联网通信, 盲信道估计, 概率图模型, 迭代接收机设计, 非正交多址接入