Preconditioning conjugate-gradient-based LAS detection for massive MIMO systems

摘要

In massive multiple-input multiple-output (MIMO) wireless systems, the computational load of data detection increases along with the increasing number of antennas. The neighbourhood search algorithms achieve near-optimal performance and these are derived from an inversion of large-dimensional matrices. Recently, linear iterative solvers, such as conjugate-gradient (CG) have been introduced to address this issue. It motivates the design of a less-complex data detection algorithm that is proficient in achieving near-optimal performance in an unconstrained ML space. Preconditioned conjugate-gradient is an approach towards further enhancement of the performance. This article proposes the computationally efficient preconditioned conjugate-gradient-based likelihood ascent search (PCGLAS) detector. PCGLAS detection algorithm achieves a fast update vector within unconstrained ML space in conjugate descent direction with little iteration. Simulation results demonstrate that the proposed algorithm can exert more influence rather than other recent state-of-the-art detection algorithms that achieve promising performance with superior running time efficiency.