LTE Downlink channel estimation based on Artificial Neural Network and complex Support Vector Machine Regression

Abstract

In this paper we assess the performance of Support Vector Machine Regression (SVR) based on Radial Basis Function (RBF) and Artificial Neural Network (ANN) based on Scaled Conjugate Gradient Backpropagation (SCG) algorithms to estimate the channel variations using the reference signal structure standardized for LTE Downlink system. Complex SVR and ANN where applied to estimate real channel environment such as vehicular A channel defined by the International Telecommunications Union (ITU). In order to evaluate the capabilities of the designed channel estimators, we provide performances of SVR and ANN, which are compared to Least Squares (LS) and Decision Feedback (DF). The simulation results show that the complex SVR has a better accuracy than other estimation techniques.