Investigations on machine learning and deep learning techniques for traffic prediction in cellular networks with big data

Abstract

The mobile cellular network offers enormous spatial and temporal newlinedata. Analysis of such volume of big data achieves cellular networks traffic newlineprediction and makes the best use of the network operators. Due to the big newlinequantity of temporal and spatial information, cellular network traffic newlineprediction is employed with some concerns. Lots of works have been newlineemployed to predict cellular network traffic. Classification is applied to newlinecategorize the input data into their corresponding classes for future prediction. newlineSimilarity-based classification is a crucial one to improve traffic prediction. newlineHowever, the prediction time and accuracy performance were not improved. newlineTo handle these concerns, cellular network traffic prediction is carried out by newlinedesigning three novel works using machine learning and deep learning newlineconcept. Novel deep learning models are used in the important similarity newlinemeasure such as Tversky similarity index and Czekanowski s dice index. This newlinesimilarity-based classification is a vital one to improve traffic prediction. newlineNovel deep learning models are employed to deeply analyze the important newlinefeatures for accurately classifying the cellular network traffic prediction. newlineFrom that, the accuracy of traffic prediction is enhanced with less time. newlineIn the first work, Deming Regression Feature Selected Modest newlineAdaptive Boosting Data Classification (DRFSMABDC) method is introduced newlineto predict the traffic in the big cellular network. The DRFSMABDC method newlineincludes three operations such as data collection, feature selection, and newlineclassification. DRFSMABDC method is designed with the novelty of the newlineDeming regression analysis, and Modest AdaBoost to enhance the accuracy newlinewith reduces the time. DRFSMABDC method initially collects the Spatiotemporal newlinedata from the input dataset. newline