Intelligent MIMO Detection Using Meta Learning

In a ${K}$ -best detector for multiple-input-multiple-output (MIMO) systems, the value of needs to be sufficiently large achieve near-maximum-likelihood (ML) performance. By treating as variable that can adjusted according fitting function some learnable coefficients, an intelligent MIMO detection network based on deep neural networks (DNN) is proposed reduce complexity algorithm with little performance degradation. particular, uses meta learning learn coefficients circumvent problem directly. The idea fusion used combine results component networks. Simulation show scheme achieves near-ML while its close linear detectors. Besides, it also exhibits strong ability fast training.