Network Traffic Prediction with Deep Learning-Based Encoder-Decoder Algorithms to Improve the Network Controller NOTED

During my summer internship at CERN, I contributed to the Network Optimised Trans- fer of Experimental Data (NOTED) project within the IT department. My work focused on enhancing the accuracy and computational efficiency of network traffic forecasting by implementing advanced Encoder-Decoder machine learning algorithms, including Seq2Seq models, Autoencoders, and Transformers. These algorithms were tested for their ability to predict network traffic and optimize data transfers across the LHCONE (Large Hadron Col- lider Open Network Environment) and LHCOPN (Large Hadron Collider Optical Private Network) links. My contributions helped improve NOTED's ability to forecast traffic more accurately and efficiently, thus supporting CERN's broader goal of optimizing data transfers for high-energy physics research.