Selection of Efficient and Accurate Prediction Algorithm for Employing Real Time 5G Data Load Prediction

In smart cities applications (i.e. intelligent transport systems, traffic management) cellular traffic load prediction is playing an essential role. The cellular data consumption can help to understand the road traffic patterns. In this context, the employment of predictive Machine Learning (ML)techniques can be useful for approximating the possible resource demands in cell towers. Therefore, cellular traffic data may very useful for finding trends and patterns of load of human activities in city traffic. In this paper, the main aim is to identify the suitable machine learning techniques, which can be used for traffic load prediction in a smart city application. The paper includes three main contributions, first providing an overview of 5G technology and their applications, second, a review on existing traffic load prediction techniques and finally, a comparative experimental study is performed among popular supervised and unsupervised learning approaches. In order to compare the performance of supervised learning algorithms, Support Vector Machine (SVM), Artificial Neural Network (ANN), Bays classifier, Linear Regression (LR), and Decision Tree (DT) are implemented. On the other hand, for comparing the performance of unsupervised learning algorithms the Self Organizing Map (SOM), Fuzzy C Means (FCM), and k-Means clustering algorithms have been involved. The experiments on publically available data on Kaggle for 4G (LTE Traffic Prediction) were used. According to the experimental analysis, SVM and ANN are accurate algorithms in the supervised learning algorithms. On the other side in unsupervised learning models, SOM shows superior accuracy. But after summarizing the results we found that the SVM and ANN algorithms are beneficial for the proposed application.