Network Traffic Prediction for 5G Network

The rapid rise of 5G technology, a smart traffic management system can benefit every customer with unlimited connections. Smart traffic detection system for predicting incoming traffic. Therefore, the load balancer can be increased/decreased according to the prediction to manage the traffic evenly and provide better Quality of services. Considering the ever-increasing load, analysis data based on traffic in the network can effectively reconfigure the structure of the multi-provider network. Traffic prediction is very important for various networks to realize resource management, predictive control, and other intelligent network operations. Forecasts for network traffic are typically expressed in terms of the volume of traffic at a particular time. It can be assumed that the current traffic conditions will continue for a while in the future if the time interval is small. This allows for the classification of the current traffic situation and the creation of a traffic forecast. We suggest a model to increase viability without sacrificing optimality in this paper. In this model a recurrent neural network is used, Gated Recurrent Unit and Long Short Term Memory achieve a balance between viability and optimality by introducing gating mechanism that allow networks to selectively update and forget information overtime there by avoiding overfitting and computational efficiency. This model shows how a costly set of measurements can be used to create a network-specific model of traffic on the network, starting with a probabilistic model of network-wide traffic. Then, rather than relying solely on inexpensive measurements, the model can be used in conjunction with them to produce predictions that are more precise. From a small number of observable links, it uses the model to forecast traffic changes on undiscovered links. Once trained, the model can be applied to a variety of network traffic cycles.