5G Network Traffic Forecast Interface (5G-NTFI): An Open Source Framework Design for Reproducibility and Replicability

Virtualization of 5G and beyond mobile networks requires high availability of computing resources to provide a reliable infrastructure. Hence, there is a need to analyze network traffic and to understand how the ability to forecast it affects the system itself and leads to more effective management of the available resources. Many theoretical and practical solutions are available to address this need, but the reproducibility of these solutions has become tedious. This work aims to provide an opensource framework, 5G Network Traffic Forecast Interface (5G-NFTI), for the reproducibility and replicability of a 5G network traffic forecast application. The implementation consists of multiple features for target group of Deep Learning (DL) experts and non-experts, allowing them to analyze and forecast 5G network traffic using DL models such as recurrent neural network (RNN), long short-term memory (LSTM), and bidirectional long short-term memory (BiLSTM). The evaluation indicates that irrespective of the chosen hardware, the framework can be replicated and consumed for testing the time-series modeling.