Research on the implementation of highly efficient MIMO equalizer for LTE-A systems based on the GPP architecture

In order to further improve the spectrum efficiency and peak data rate, uplink MIMO (multiple - input and multiple-output) has been introduced into LTEA (Release 10 and beyond). As the large dimensional MIMO increases system complexity, it is a challenge for the implementation of real-time MIMO equalizer. The massive floating-point computing power and flexible multi-core parallelism of the GPP (general purpose processor) platform make it possible to implement highly efficient MIMO equalizer. In this paper, we present our initial view on the optimization of MIMO equalizer based on the GPP platform. Furthermore, we analyze the principle of the improved matrix inversion algorithm based on Cholesky decomposition, and explain the ways to reduce computational complexity using SIMD instructions and multi-thread technology.