Machine Learning Model of Communication of Physical and Virtual Sensors in the Mobile Network on the Motorway Section

The effects of the communication of physical and virtual sensors on the performance of the mobile network can be assessed through several Key Performance Indicators (KPI), of which the Average Downlink Throughput (ADT) is very important. The paper presents an approach for measuring this indicator in a mobile network using a Machine Learning Model (MLM) which uses contextual variables related to traffic class, user location, time period, signal strength, base station. The geographical area of the m:tel network on the section of the “9th January” motorway, a key road in the Republic of Srpska, BiH, was selected for the case study in the research. The measured ADT performance reflects the degree of Quality of customer Service (QoS) in the observed space. The proposed model was created in the SPSS software package, and its training and testing were performed based on data collected by field measurements. The research results show that the model can perform measurement and prediction with satisfactory accuracy while increasing the data set for training and testing by the Data Augmentation (DA) method. In addition to the MLM model, the paper proposes a wireless sensor network model for the acquisition of contextual data along the motorway.