Deep Learning Based Pilot Allocation Scheme (DL-PAS) for 5G Massive MIMO System
Deep Learning Based Pilot Allocation Scheme (DL-PAS) for 5G Massive MIMO System
复制标题
DOI：
10.1109/lcomm.2018.2803054
复制
发表时间：
2018-04-01
期刊：
影响因子：
--
通讯作者：
Choi, Junkyun
中科院分区：
文献类型：
--
作者：
Kim, Kwihoon;Lee, Joohyung;Choi, Junkyun
本文提出了一种基于深度学习的导频分配方案（DL-PAS），用于大规模多输入多输出（massive MIMO）系统，该系统为多个用户使用大量天线。提出的DL-PAS通过学习导频分配与用户位置模式之间的关系，提高了导频污染严重的蜂窝网络中的性能。在这封信中，我们设计了一种新的监督学习方法，其中输入特征和输出标签分别是用户在所有单元和导频分配中的位置。具体来说，以给定用户位置为训练数据，通过穷举搜索方法提供预训练的最优飞行员分配。然后，本文提出的DL-PAS通过分析训练数据，从生成的推断函数中提供接近最优的飞行员分配。我们使用商业深度多层感知器系统来实现所提出的方案。仿真实验表明，该方案在较低的复杂度下实现了99.38%的理论上限性能，计算时间仅为0.92 ms。
This letter proposes a deep learning-based pilot assignment scheme (DL-PAS) for a massive multiple-input multiple-output (massive MIMO) system that utilizes a large number of antennas for multiple users. The proposed DL-PAS improves the performance in cellular networks with severe pilot contamination by learning the relationship between pilot assignment and the users' location pattern. In this letter, we design a novel supervised learning method, where input features and output labels are users' locations in all cells and pilot assignments, respectively. Specifically, pretrained optimal pilot assignments with given users' locations are provided through an exhaustive search method as the training data. Then, the proposed DL-PAS provides a near-optimal pilot assignment from the produced inferred function by analyzing the training data. We implement the proposed scheme using a commercial deep multilayer perceptron system. Simulation-based experiments show that the proposed scheme achieves almost 99.38% theoretical upper-bound performance with low complexity, requiring only 0.92-ms computational time.