Machine Learning Enabled Performance Prediction Model for Massive-MIMO HetNet System

• Main Authors: Shuvabrata Bandopadhaya, Soumya Ranjan Samal, Vladimir Poulkov
• Format: Article
• Language: English
• Published: MDPI AG 2021-01-01
• Series: Sensors
• Subjects: 5G; B5G wireless networks; massive MIMO; HetNet; machine learning; coverage probability
• Online Access: https://www.mdpi.com/1424-8220/21/3/800

To support upcoming novel applications, fifth generation (5G) and beyond 5G (B5G) wireless networks are being propelled to deploy an ultra-dense network with an ultra-high spectral efficiency using the combination of heterogeneous network (HetNet) solutions and massive Multiple Input Multiple Output (MIMO). As the deployment of massive MIMO HetNet systems involves a high capital expenditure, network service providers need a precise performance analysis before investment. The performance of such networks is limited because of presence of inter-cell and inter-tier interferences. The conventional analytic approach to model the performance of such networks is not trivial, as the performance is a stochastic function of many network parameters. This paper proposes a machine learning (ML) approach to predict the network performance of a massive MIMO HetNet system considering a multi-cell scenario. This paper considers a two-tier network in which the base stations of each tier are equipped with massive MIMO systems working in a sub 6GHz band. The coverage probability (CP) and area spectral efficiency (ASE) are considered to be the network performance metrics that quantify the reliability and achievable rate in the network, respectively. Here, an ML model is inferred to predict the numerical values of the performance metrics for an arbitrary network configuration. In the process of practical deployments of future networks, the use of this model could be very valuable.