National Technical University of Athens (NTUA)
Δυναμική διαχείριση πόρων οπισθοζευκτικού δικτύου σε δίκτυα κινητών επικοινωνιών νέας γενιάς
Abstract
dc:descriptionThe tremendous growth of the mobile users’ needs has paved the way for the introduction of the next generation mobile networks. As a result, mobile operators are forced to adopt the new communication standards in order to be able to satisfy their subscribers’ needs. However, the high information rates that should be offered by the wireless access network can potentially create a bottleneck at the backhaul network, which was traditionally designed to support voice services. In order to overcome such situation, the convergence of the wireless access networks with a passive optical network at the backhaul has been proposed in the literature. The extremely high data rates that can be offered by the optical backhauling network can achieve the requirements of the 4G standard providing satisfactory Quality of Service (QoS) to the end-users. However, in this case there are many issues that have to be taken into consideration for the efficient allocation of the resources in the network. In particular, traditional methods of network planning, which are based on a worst case scenario, lead to a static and inefficient resource allocation, and thus, they cannot be implemented within a converged optical-wireless network.In this context, this thesis studies the problem of the dynamic resource allocation at the backhaul network and proposes a distributed system for dynamic resource management that can address the needs of the next generation mobile networks. In particular, an intelligent agent, which uses machine learning techniques in order to estimate the forthcoming demand at the Base Station and proactively request the commitment of the appropriate resources, is developed. Specifically, the proposed system monitors and collects the necessary traffic statistics of the backhaul network, and based on the collected data, it calculates the appropriate resources in order to create a service level agreement between the mobile operator and the operator of the passive optical network.All the appropriate functions that allow the dynamic resource management are described. In particular, more effort is given to the forecasting process, which consist the basic functionality of the proposed system. Specifically, the performance of different machine learning techniques is investigated and issues that contribute to the optimization of the system’s functionality are studied both at a practical and a theoretical level. Furthermore, apart from the forecasting process, which aims to the creation of a service level agreement, the short-term forecasting process is also studied. It uses a combination of machine learning techniques and time series analysis and its scope is to address the unexpected variations in the network traffic. Based on the forecasting results, useful conclusions about the operation and the performance of the system are extracted.Finally, the resource management problem at the backhaul network is studied using evolutionary game theory. Specifically, the problem is studied at the side of the Base Station, where the interactions between the Base Station and its subscribers are properly modeled, and at the side of the optical network, where different mobile operators share a common backhaul infrastructure and compete for its resources. The interactions between the players are modeled using replicator dynamics and the stability of the system is proven. The proposed schemes are validated through simulation results.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2014
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Loumiotis, Ioannis
- Λουμιώτης, Ιωάννης
Subjects
dc:subject × 30- Οπισθοζευκτικό δίκτυο
- Δυναμική διαχείριση πόρων
- Δίκτυα νέας γενιάς
- Παθητικό οπτικό δίκτυο
- Πρόβλεψη
- Τεχνητά νευρωνικά δίκτυα
- Πολυστρωματικά δίκτυα
- Νευρωνικά δίκτυα γενικευμένης παλινδρόμησης
- Νευρωνικά δίκτυα ομαδικής διαχείρισης δεδομένων
- Εξελικτική θεωρία παιγνίων
- Ανάλυση διακλάδωσης
- Δυναμική των αντιγραφέων
- Καθυστέρηση
- Backhaul network
- Dynamic resource management
- Next generation networks
- Passive optical network
- Forecasting
- Artificial neural networks
- Multilayer perceptrons
- Gegeneral regression neural networks
- Group method of data handling neural network
- Evolutionary game theory
- Bifurcation analysis
- Replicator dynamics
- Delay
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Engineering and Technology
- Electrical Engineering, Electronic Engineering, Information Engineering
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/39571