基于深度神经网络的OFDM系统IQ不平衡补偿技术

正交频分复用(OFDM)是现代移动通信中一项重要的物理层通信技术，并且OFDM系统要求子载波间严格正交。然而在实际系统中，振荡器和滤波器等器件的非理想特性会导致同相正交(In-phase and Quadrature-phase，IQ)不平衡，从而破坏子载波的正交性，严重影响OFDM系统的性能。本文研究了IQ不平衡对OFDM系统的影响，提出了基于并联深度神经网络的IQ不平衡补偿技术。该算法利用了深度神经网络不依赖于模型的特点，直接从接收到的频域信号恢复原输入信号的二进制序列，并利用IQ不平衡的干扰信号来自镜像子载波的先验知识来初始化神经网络，加快其网络优化的收敛速度。仿真结果表明，该算法能有效地补偿IQ不平衡失真，并且在幅度和相位失真的补偿上，其性能都优于传统的导频的最小二乘（LS）估计补偿IQ不平衡的算法，证明了深度学习方法解决物理层问题的优越性。

IQ Imbalance Compensation Based on Deep Neural Network in OFDM Systems

Orthogonal frequency division multiplexing (OFDM) is an essential technique in the physical layer of wireless communications, and OFDM system requires rigid orthogonality between subcarriers. However, in practical systems, the imperfect of components like the oscillator and filter would introduce IQ imbalance into the system. IQ imbalance would interfere the orthogonality between subcarriers and decrease the system performance. This article discusses the effect of IQ imbalance and proposes an IQ imbalance compensation technique based on parallel deep neural network (DNN). The deep neural network relies rarely on mathematic models, and our algorithm utilizes this feature to recover the original signal from the received signal in the frequency domain to its original binary sequence of transmitted signal directly. Meanwhile, we utilize the expert knowledge that the IQ imbalance interference comes from the image aliasing effect to initialize the network. Simulation results have proved our algorithm can effectively compensate IQ imbalance distortion, and it outperforms traditional LS algorithm based on pilots in both amplitude and phase compensation and proves the superiority of deep learning solutions for issues in the physical layer