Comparison of Machine Learning Methods for IoT and IIoT Traffic Prediction

Recently, artificial intelligence have demonstrated significant success in solving the problem of analyzing and forecasting both telecommunication traffic and Internet of Things traffic. Using the example of data collected from mobile devices installed in the intelligent transportation system, an overview and analysis of various time series forecasting models describing the behavior of IoT traffic are presented. Such forecasting models as the exponential smoothing method, linear regression, autoregressive integrated moving average (ARIMA) method, support vector machine regression method, N-BEATS method using fully connected layers of a neural network for forecasting one-dimensional time series are discussed. Features of some of them are briefly described. The data preparation operations are described for a specific data array: removing unused columns, replacing missing data on transaction duration with their median values. The main statistical characteristics of the data array are described. A preliminary data analysis is presented, which consists of applying smoothing methods: moving average and exponential smoothing. The process of training models and a comparative analysis of their training quality are described. It has been concluded that for the UDP protocol, the ARIMA model has the best training quality, for the TCP protocol - linear regression and the Theta model, for the HTTPS protocol - linear regression, ARIMA and N-BEATS.