A Low Complexity Expectation Propagation Detector for Extra-Large Scale Massive MIMO

In this paper, we propose a low-complexity expectation propagation (EP) detector for extra-large scale massive multiple-input multiple-output (MIMO) systems. EP achieves the near-optimal performance in high-dimensional MIMO systems but suffers from the inevitable cost of computational complexity. In each iteration procedure of EP algorithm, matrix inversion is the main computational burden, which is the key challenge of the practical implementation of EP detectors. In this work, we use polynomial expansion (PE) to approximate matrix inverse, which reduces the computational cost of matrix inverse in each EP iteration. A subarray architecture is considered in our scenario to fit extra-large scale antenna array. Numerical analysis shows that the proposed PE-EP detector has extremely low complexity for extra-large scale antenna array and achieves performance similar to MMSE and original EP, which ensures enhanced performance and complexity trade-off.