Ερευνητικό Προσωπικό
Πλαστράς Στέφανος
Προσωπικά Στοιχεία
Πλαστράς Στέφανος
s [dot] plastras [at] aegean [dot] gr
Γραφείο Β9, Κτήριο Λυμπέρη, Οδός Παλαμά 2, Νέο Καρλόβασι, Σάμος, GR 83200
Κατόπιν επικοινωνίας
Copyright Notice: Το υλικό αυτό παρουσιάζεται για έγκαιρη διάδοση επιστημονικής και τεχνικής εργασίας. Τα πνευματικά δικαιώματα και όλα τα σχετικά δικαιώματα παραμένουν στους συγγραφείς ή σε άλλους κατόχους πνευματικών δικαιωμάτων. Όσοι αντιγράφουν αυτές τις πληροφορίες αναμένεται να τηρούν τους όρους και τους περιορισμούς που επιβάλλει το πνευματικό δικαίωμα κάθε συγγραφέα. Στις περισσότερες περιπτώσεις, τα έργα αυτά δεν μπορούν να αναδημοσιευτούν ή να αναπαραχθούν μαζικά χωρίς τη ρητή άδεια του κατόχου των πνευματικών δικαιωμάτων.
Επιστημονικά Συνέδρια
E. Oikonomou, S.Plastras, D. Tsoumatidis, D. N. Skoutas, A. Rouskas, Workload Prediction for Efficient Node Management in Mobile Edge Computing, 2024 IFIP Networking Conference (IFIP Networking), Aug, 2024, Thessaloniki, Greece , IEEE, https://doi.org/10.23919/IFIPNetworking6...
Περίληψη:
The rapidly increasing number of mobile devices and resource-intensive applications poses substantial obstacles to traditional centralized mobile cloud computing, resulting in increased latency and decreased service quality. Edge computing, which places server capabilities at access nodes, provides a promising solution to the aforementioned issues. However, maintaining operational edge service nodes at each access node can be costly and inefficient. We propose and evaluate a scheme that combines a heuristic service node selection algorithm with machine learning based computational load prediction, with the goal of minimizing latency and balancing load among service nodes. Simulation experiments demonstrate that the proposed scheme substantially enhances system performance, paving the way for a more efficient and responsive edge network infrastructure, particularly in 5G and 6G mobile communication environments.
The rapidly increasing number of mobile devices and resource-intensive applications poses substantial obstacles to traditional centralized mobile cloud computing, resulting in increased latency and decreased service quality. Edge computing, which places server capabilities at access nodes, provides a promising solution to the aforementioned issues. However, maintaining operational edge service nodes at each access node can be costly and inefficient. We propose and evaluate a scheme that combines a heuristic service node selection algorithm with machine learning based computational load prediction, with the goal of minimizing latency and balancing load among service nodes. Simulation experiments demonstrate that the proposed scheme substantially enhances system performance, paving the way for a more efficient and responsive edge network infrastructure, particularly in 5G and 6G mobile communication environments.