基于硬件损伤和非完美CSI的IRS辅助NOMA网络鲁棒传输算法

Robust transmission algorithm for IRS-assisted NOMA network with hardware impairments and imperfect CSI

为提升非正交多址接入(NOMA,non-orthogonal multiple access)网络的鲁棒性并减少其能量消耗,考虑了收发器的硬件损伤(HWI,hardware impairment)和非完美的信道状态信息(CSI,channel state information),提出了智能反射面(IRS,intelligent reflecting surface)辅助的NOMA网络传输功率最小化算法。考虑了用户服务质量(QoS,quality of service)约束、串行干扰消除约束以及IRS的反射相移约束,基于HWI和非完美CSI建立了基站主动波束成形和IRS被动波束成形联合优化问题模型。为求解该非凸优化问题,首先利用线性近似和S-Procedure方法对QoS约束进行转换,然后将优化问题分解为两个子问题,并且利用逐次凸逼近(SCA,successive convex approximation)方法求解主动波束成形子问题,利用惩罚凸凹过程算法求解被动波束成形子问题,最后利用交替优化将子问题交替迭代得到最终解。仿真结果表明,在传输功率方面,所提算法比正交多址鲁棒算法降低了17.05%;在系统鲁棒性方面,所提算法相较于HWI鲁棒算法和CSI鲁棒算法,分别提升了20.69%和31.14%。

To improve the robustness and reduce the energy consumption of non-orthogonal multiple access(NOMA)networks,based on hardware impairments(HWI)of transceiver and non-perfect channel state information(CSI),an intelligent reflecting surface(IRS)assisted transmission power minimization algorithm for NOMA networks was proposed.The joint optimization problem of active beam assignment at the base station and passive beam assignment at the IRS was modeled based on HWI and non-perfect CSI.The system considered the user quality of service(QoS)constraint,the serial interference cancellation constraint and the reflection phase shift constraint of the IRS.To solve this nonconvex optimization problem,the QoS constraints were firstly transformed using linear approximation and S-Procedure methods.Then the optimization problem was decomposed into two subproblems.The active beam assignment subproblem was solved using the successive convex approximation(SCA)method.The passive beam assignment subproblem was solved using the penalized convex-concave process algorithm.Finally,the final solution was obtained by iterating the subproblems alternately using alternating optimization.The simulation results show that the proposed algorithm reduces 17.05%compared to the orthogonal multiple access robust algorithm in terms of transmitted power.In terms of system robustness,the proposed algorithm improves by 20.69%and 31.14%compared to the HWI robust algorithm and the CSI robust algorithm,respectively.