Short CV: Christos Bouras is Professor in the University of Patras, Department of Computer Engineering and Informatics. Also he is a scientific advisor of Research Unit 6 in Computer Technology Institute and Press – Diophantus, Patras, Greece. His research interests include Analysis of Performance of Networking and Computer Systems, Computer Networks and Protocols, Mobile and Wireless Communications, Telematics and New Services, QoS and Pricing for Networks and Services, e-learning, Networked Virtual Environments and WWW Issues. He has extended professional experience in Design and Analysis of Networks, Protocols, Telematics and New Services. He has published more than 450 papers in various well-known refereed books, conferences and journals. He is a co-author of 9 books in Greek and editor of 2 in English. He has been member of editorial board for international journals and PC member and referee in various international journals and conferences. He has participated in R&D projects.
@article{4547,
title = {Prediction Mechanisms to Improve 5G Network User allocation and Resource management},
author = {Christos Bouras and Rafail Kalogeropoulos},
url = {https://telematics.upatras.gr/bouras-kalogeropoulos2022_article_predictionmechanismstoimprove5/},
year = {2022},
date = {2022-01-01},
urldate = {2022-01-01},
journal = {Wireless Personal Communications, Springer Verlang},
pages = {1 - 25},
abstract = {As technology rapidly advances, the number of devices constantly communicating, transmitting and receiving data through the cellular networks keeps rising, posing an unprecedented load on them. Such an increase calls for establishing new methods to manage these devices as well as utilize the data produced by them to establish network architectures that can prevent cellular networks from overloading. To achieve the desired results, we need to optimally allocate network resources to existing users. Resource allocation has traditionally been considered an optimization problem where proposed solutions are hard to implement in real time, resulting in the use of inferior solutions with reduced produced performance. With the introduction of Machine Learning, we propose three mechanisms, intending to utilize network data to improve real time network performance. The first mechanism, a Decision Trees based mechanism aims to improve real time decision making by predicting the optimal matching of users and Base Stations. The second mechanism, a K-means based mechanism intends to tackle network congestion and ensure uninterrupted Quality of Service by predicting the optimal coordinates for placing Base Stations along the network based on traffic data. Finally, a Regression based mechanism manages to predict user movement along the network, resulting in improved resource management and reduced energy waste. These mechanisms can work cooperatively, essentially establishing a network architecture that utilizes prediction to efficiently allocate users and manage available resources.},
keywords = {Internet of Things, Machine Learning},
pubstate = {published},
tppubtype = {article}
}
As technology rapidly advances, the number of devices constantly communicating, transmitting and receiving data through the cellular networks keeps rising, posing an unprecedented load on them. Such an increase calls for establishing new methods to manage these devices as well as utilize the data produced by them to establish network architectures that can prevent cellular networks from overloading. To achieve the desired results, we need to optimally allocate network resources to existing users. Resource allocation has traditionally been considered an optimization problem where proposed solutions are hard to implement in real time, resulting in the use of inferior solutions with reduced produced performance. With the introduction of Machine Learning, we propose three mechanisms, intending to utilize network data to improve real time network performance. The first mechanism, a Decision Trees based mechanism aims to improve real time decision making by predicting the optimal matching of users and Base Stations. The second mechanism, a K-means based mechanism intends to tackle network congestion and ensure uninterrupted Quality of Service by predicting the optimal coordinates for placing Base Stations along the network based on traffic data. Finally, a Regression based mechanism manages to predict user movement along the network, resulting in improved resource management and reduced energy waste. These mechanisms can work cooperatively, essentially establishing a network architecture that utilizes prediction to efficiently allocate users and manage available resources.
