Towards Software-Defined Massive MIMO for 5G&B Spectral-Efficient Networks

Abstract

As a key enabling technology for 5G&B systems, massive multi-input multi-output (MIMO) allows the system capacity to be theoretically increased by simply installing additional antennas to remote radio heads (RRHs). However, this innovative technology cannot support higher data capacity without accurate channel state information and interference handling, especially for multi-cell scenarios. In this paper, the dynamic macro- diversity (i.e., network) massive MIMO is treated from the perspective of software-defined cellular architecture. The so-called software-defined massive MIMO is introduced, which dynamically coordinates highly-deployed RRHs equipped with massive antennas so that the maximum spectral efficiency is achieved. First, the software- defined cellular architecture is presented, where distributed massive antenna systems with centralized control and time-division duplexing massive MIMO are investigated. Next, in this considered architecture, a rigid analysis of achievable ergodic user sum-rates is given for macro-diversity massive MIMO schemes. An optimization framework of software-defined massive MIMO is further proposed that optimizes RRH clustering pattern and RRH-user associations while satisfying system-level constraints. To address the NP-complete problem of the optimal framework design, an iterative, global search algorithm is developed that exploits genetic algorithms and yields satisfactory solutions in only few rounds. Performance evaluation validates the efficacy of our solution which facilitates universal frequency reuse for 5G&B wireless networks.