Volume 18, No. 6, 2021
Cellular Traffic Prediction Using Deep Learning-Based Novel Fusion Neural Network And Traffic Variation Handling Algorithm
Supriya H.S , Dr. Chandrakala B.M
Abstract
Big data constitutes a huge amount of data that can be in form of structured or unstructured, these data are stored on daily basis. Such large volume data requires an efficient processing mechanism as it is considered as challenging as well as a complex task. However, these data are often helpful in prediction, one of the applications remains in traffic prediction in cellular networks. Traffic load prediction on network links helps in designing the resource allocation strategies, hence several types of research have been carried out in past for traffic prediction including the deep learningbased model and deep learning-based architecture are found to be more efficient than any traditional mechanism. Hence this research adopts the two different neural networks and designed Fusion Neural Network to enhance the prediction. FNN is an integration of CNN and RNN and its layer; this novel FNN architecture not only enables the exploitation of the topological properties and aims to predict the load link on the network-based but also captures and exploits custom features that establish the link relationships in the network After designing an architecture, a novel Traffic variation handling algorithm is designed to optimal error prediction. FNN is evaluated considering the Telecom Italia Big Data Challenge dataset over the different the metrics like RMSE and MAE considering the SMS, call and internet service; further evaluation is carried out through comparing with an existing model and comparative analysis proves it.
Pages: 1976-1994
Keywords: The growing advancements in technology have resulted in the evolution of smartphones over the decade. Which has led to a rapid explosion and generation of data, pacing up the big data era [1].