Energy Efficient Power Allocation for Downlink NOMA Heterogeneous Networks With Imperfect CSI

• Published in: IEEE access Vol. 7; pp. 39329 - 39340
• Main Authors: Song, Xin, Dong, Li, Wang, Jingpu, Qin, Lei, Han, Xiuwei
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2019; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Computational geometry; Convexity; Downlinking; Energy efficiency; Fading channels; Heterogeneous networks; imperfect channel state information; Interference; Mathematical programming; New technology; NOMA; non-orthogonal multiple access; Nonorthogonal multiple access; Optimization; power allocation; Resource management; Search algorithms; Statistical analysis; Wireless networks
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2019.2906780

Non-orthogonal multiple access (NOMA) is a promising emerging technology that can significantly improve the utilization of spectrum and system capacity in heterogeneous wireless networks. Power allocation plays a key role in the successful deployment of NOMA. In the most prior power allocation schemes, the perfect channel state information (CSI) is assumed to be known which is difficult to obtain in a realistic environment. In this paper, we propose a power allocation scheme to maximize energy efficiency for downlink NOMA heterogeneous networks based on imperfect CSI. The system model for imperfect CSI is built, in which the optimization problem is a probabilistic non-convex problem with the constraint of outage probability. To solve the optimization problem, the probabilistic problem is transformed to a non-probabilistic problem through relaxation. The power allocation for each small cell is achieved via bisection search algorithm based on gradient value, where the trend of energy efficiency as a function of the power of the small cell is analyzed. The sequential convex programming is adapted to transform the non-convex problem to a convex problem. The closed-form solutions of power allocation factors are derived by the Lagrangian multiplier method. The simulation results show the superiority and efficiency of the proposed scheme compared with the traditional algorithms.