Analysis of Channel Estimation using Machine Learning Algorithms for Next Generation Wireless Communication

Abstract

The high-end data-oriented multiple input and multiple output system using conventional channel estimation compromised with performance and reliability of service. Moreover, the ageing impact of various factors such as time, frequency, propagation and multipath degraded the performance cannot meet the level of the next-generation wireless communication system. To resolve the problem of noise and services, the channel estimation methods move on to the area of neural networks and machine learning. The efficiency and utility of machine learning (ML) enrich the capacity of channel estimation in all network scenarios. The neural network-based model increases the performance of least-square (LS) and LMMSE channel estimation like linear and stationary.  The switching of non-linear and non-stationary the efficiency of the neural network model is compromised, but specific models work efficiently, such as RNN and CNN. This paper presents the compressive study of channel estimation and analysis of machine learning-based channel estimation methods. The analysis of channel estimation applies to the MIMO system in different fading conditions.  The machine learning algorithm can learn channel structure and estimate channels from many training data. Furthermore, we analyse the performance of different ML algorithms of varying data sizes. Based on our analysis and simulation results, machine learning significantly outperforms in MIMO systems