Machine Learning Based Link-to-System Mapping for System-Level Simulation of Cellular Networks

Abstract

This paper proposes a machine learning (ML)-based exponential effective signal-to-noise ratio (SNR) mapping (EESM) method for simulating the system-level performance of cellular networks, which utilizes a deep neural network (DNN) regression algorithm. We first explain overall procedure of the link-to-system (L2S) mapping algorithm which has been used in commercial standardization organizations such as IEEE 802.16 and 3GPP LTE. Then, we apply the proposed ML-based EESM method to the existing L2S mapping procedure. The processing time of the L2S mapping becomes significantly reduced through the proposed method while the mean squared errors (MSE) between the actual block-error rate (BLER) from the link-level simulator and the estimated BLER from the L2S mapping technique is also decreased, compared with the conventional L2S mapping method.