Max-Min QoS Power Control in Generalized Cell-Free Massive MIMO-NOMA With Optimal Backhaul Combining

• Published in: IEEE transactions on vehicular technology 2020-10, Vol.69 (10), p.10949-10964
• Main Authors: Nguyen, The Khai, Nguyen, Ha H., Tuan, Hoang Duong
• Format: Article
• Language: English
• Subjects: Algorithms; Antenna arrays; Antennas; cell-free networks; Central processing units; CPUs; Diversity reception; energy efficiency; Interference; Massive MIMO; Mathematical analysis; MIMO (control systems); NOMA; Nonorthogonal multiple access; optimal combining; Optimization; Power control; Sequences; Service areas; Signal to noise ratio; spectrum efficiency; Stations
• ISSN: 0018-9545; 1939-9359
• DOI: 10.1109/TVT.2020.3006054

This paper studies the uplink (UL) transmission of a generalized cell-free massive multiple-input multiple-output (massive MIMO) system in which multiple base stations (or access points), each equipped with a multiple-antenna array and connected to a central processing unit (CPU) over a backhaul network, simultaneously serve multiple users in a cell-free service area. The paper focuses on the non-orthogonal multiple access (NOMA) approach for sharing pilot sequences among users. Unlike the conventional cell-free massive MIMO-NOMA systems in which the UL signals from different access points are equally combined over the backhaul network, this paper first develops an optimal backhaul combining (OBC) method to maximize the UL signal-to-interference-plus-noise ratio (SINR). It is shown that, by using OBC, the correlated interference can be effectively mitigated if the number of users assigned to each pilot sequence is less than or equal to the number of base stations (BSs). As a result, the cell-free massive MIMO-NOMA system with OBC can enjoy unlimited performance when the number of antennas at each BS tends to infinity. A closed-form SINR expression is derived under Rayleigh fading and used to formulate a max-min quality-of-service (QoS) power control problem to further enhance the system performance. To deal with the NP-hardness of the concerned optimization problem, a successive inner approximation technique is applied to convert the original problem into a series of convex optimizations, which can be solved iteratively. In addition, a user grouping algorithm is also developed and shown to be better than random user grouping and a grouping method recently proposed in the literature. Numerical results are presented to corroborate the analysis and demonstrate the superiority of the proposed optimal backhaul combining over both equal-gain backhaul combining and zero-forcing backhaul combining.