Machine Learning Based Link Adaptation Method for MIMO System

Link Adaptation can maximize system throughput while maintaining transmission reliability. With the growing demand for high-speed data transmission, multiple-input multiple-output (MIMO) and orthogonal frequency division multiplexing (OFDM) technologies have been widely used in wireless communication systems. However, performing link adaptation in MIMO systems is challenging due to the complexity of channel and coupling among equalization, precoding, spatial mode, modulation and coding scheme (MCS). In this paper, we present a link adaptation scheme in MIMO systems through machine learning algorithms to maximize spectral efficiency while maintaining transmission reliability. We propose to use autoencoder model to extract feature from channel state information (CSI), combined with logical regression algorithms to select modulation and coding scheme. Spatial mode can be chosen based on the objective of maximizing the spectral efficiency. Simulation results demonstrate the improved performance and validate the application of the proposed learning based framework in MIMO systems.