A Review of Coverage Predictions in Mobile Communication Systems using Machine Learning Algorithm

The needs for best signal coverage and high performance cellular network quality is critical in planning, designing, and operating 5G cellular networks due to the maturity of Internet penetration in today’s society. In the process of cellular network planning and optimization, accurate predictions are required. With this accurate coverage prediction, mobile network operators can identify areas that require network infrastructure improvements, as well as optimize network resources to improve the quality of service provided. There are several factors that influence coverage analysis and prediction, including the distance between the transmitter and receiver, topographic depiction of the area modeled in the propagation path loss model, weather, signal interference, and other factors. There are several methods for predicting coverage or coverage of signal power levels at the receiver. These methods have been developed in previous research including model-based prediction methods and data-driven prediction methods. Model-based prediction methods are based on empirical models of network characteristics and environmental parameters, while data-driven prediction methods are based on processing signal data and environmental data that have been collected from cellular networks and simulated using machine learning algorithms. In addition to these two methods, there are also studies that use a combination of the two methods to be used in predicting coverage or predicting other functions in cellular communication systems. This paper will present a review of research papers related to these two methods, including the development of the methods.