Optimizing Weighted-Sum Energy Efficiency in Downlink and Uplink NOMA Systems

• Published in: IEEE transactions on vehicular technology Vol. 69; no. 10; pp. 11112 - 11127
• Main Authors: Zamani, Mohammad Reza, Eslami, Mohsen, Khorramizadeh, Mostafa, Zamani, Hojatollah, Ding, Zhiguo
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.10.2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Computational geometry; Convexity; Dinkelbach-like; Downlink; Downlinking; Energy conversion efficiency; Energy efficiency; Linear functions; Measurement; NOMA; Non-orthogonal multiple access (NOMA); Nonorthogonal multiple access; Optimization; Power consumption; power consumption model (PCM); Power demand; Resource management; Signal processing; successive convex approximation (SCA); Uplink; weighted sum energy efficinecy (WSEE)
• ISSN: 0018-9545; 1939-9359
• DOI: 10.1109/TVT.2020.3007716

In this paper, weighted sum energy efficiency (WSEE) in uplink and downlink of a multi-user non-orthogonal multiple access (NOMA) system is considered. we adopt a more realistic power consumption model where signal processing power is modeled as a linear function of transmit power. Rather than the well-known network-centric global energy efficiency (GEE) metric, which is a pseudo-concave (PC) function, the weighted sum energy efficiency metric is considered, which is not PC in general. To find the optimum user power allocation Dinkelbach-like algorithm is adopted, by which each individual fractional EE function is converted to a parametric function where under some conditions on the weights falls into a class of convex optimization problems and it is solved in the dual domain. The dual variables are updated using a sub-gradient and cutting plane-based algorithm, which here ellipsoid method is used. Since the optimum solution restricts user weights, a low complexity suboptimum algorithm that does not consider any condition on user's weights is proposed. The problem is non-convex in general; hence, epigraph form followed by successive convex approximation (SCA) is used to deal with that problem. Results demonstrate that with the user-oriented metric, one can provide different priorities to users, and by choosing proper weights entail fairness among users.