National and Kapodistrian University of Athens
Context-based resource management and slicing for SDN-enabled 5G smart, connected environments
Abstract
dc:descriptionThe fifth-generation (5G) mobile communication systems, which are expected to emerge in the forthcoming years, will address unprecedented demands in terms of system capacity, service latency and number of connected devices. Future 5G network ecosystems will comprise a plethora of 3GPP and non-3GGP Radio Access Technologies (RATs), such as Wi-Fi, 3G, 4G or LTE, Bluetooth, etc. Deployment scenarios envision a multi-layer combination of macro, micro and femto cells where multi-mode end devices, supporting diverse applications, are served by different technologies. Limitations previously posed by legacy generation systems need to be eliminated, paving the way to a new wave of services and overall experience for the user. As a result, the management of radio resources via mapping the end devices to the most appropriate access network becomes of paramount importance; the primary Radio Resource Management (RRM) mechanisms, i.e. cell selection/reselection, handover and call admission control will be able to offer extremely high Quality of Service (QoS) and Experience (QoE) to the users, towards the very demanding 5G use case requirements; this will be realised via an optimal association between the diverse end devices and the coexisting available access networks. Besides the user’s perspective, the Mobile Network Operators (MNOs) will be able to take advantage of the maximum efficiency and utilization over the –already scarce- wireless resources. Intelligent optimizations, as well as cost and energy efficient solutions need to be introduced in 5G networks in order to promote a consistent, user-centred and all-dimensional information ecosystem.This thesis primarily focuses on the radio resource management (RRM) from the perspective of the primary RAT and cell layer selection processes (i.e., cell (re)selection, handover, admission control); afterwards, it goes one step beyond, in order to link the RRM with one of the latest RRM optimization approaches, i.e. the Network Slicing, as introduced in Software Defined Networking (SDN)-enabled environments, which creates smaller, virtual “portions” of the network, adapted and optimized for specific services/requirements. As a first step, a comprehensive analysis for the existing solutions -as these are specified in 3GPP standards, research papers, and patents has taken place. This thesis initially identifies the links between the research community efforts, the industry implementations, as well as the standardization efforts, in an attempt to highlight realistic solution implementations, identify the main goals, advantages and shortcomings of these efforts. As will be shown, existing solutions attempt to balance between implementation simplicity and solution optimality. Thus, solutions are either simple to implement but achieve sub-optimal solutions or provide significant improvements but their complexity and the burden placed on the network components –in terms of processing, as well as signaling resources- renders them unattractive for a real-life deployment.Towards this end, this thesis introduces a context-based radio resource management (RRM) framework, comprised of three distinct mechanisms: Two out of the three mechanisms exploit contextual information with the aim of optimising the resource management and UE flows-RAT mapping, while the third mechanism acts with an augmenting role to the former two, by pre-processing the contextual information required by such, context-based mechanisms and –thus- by limiting the signaling cost required for communicating this contextual information among network entities. In addition to the three mechanisms, comprehensive analysis has taken place in relation to architectural aspects, in the context of the forthcoming 5G network architecture and by mapping them with the latest 5G network components –as these were introduced in the latest 3GPP work-.The first major contribution of this thesis is COmpAsS, a context-aware, multi-criteria RAT selection mechanism, the main part of which operates on the User Equipment (UE) side, minimizing signaling overhead over the air interface and computation load on the base stations. COmpAsS mechanism performs real-time monitoring and exploits the Fuzzy Logic (FL) approach as the core logic component, responsible for the perception of the network situation and -in combination with a set of pre-defined rules, calculates a list of the most suitable available access network options, for each one of the UE’s active data flows/services. The merits of COmpAsS are showcased via an extensive series of simulation scenarios, as part of 5G ultra dense networks (UDN) use cases. The results prove how the proposed mechanism optimises Key Performance Indicators (KPIs), when juxtaposed to a well-established LTE handover algorithm. The second major contribution of the current thesis the Context Extraction and Profiling Engine (CEPE), a resource management framework, which analyzes user behavioral patterns, extracts meaningful knowledge and performs user profiling in order to apply it for optimal resource planning, as well as prediction of resource requirements. CEPE collects information about users, services, terminals and network conditions and –based on offline processing– derives a knowledge model, which is subsequently used for the optimization of the primary RRM mechanism. Then, the extracted context information is translated into user profiles and is finally applied as input for enhanced cell (re)selection, handover or admission control. The viability and validity of CEPE is demonstrated via an extensive set of simulation scenarios.The third major contribution is CIP, a Context Information Pre-processing scheme, aiming to identify and discard redundant or unnecessary data prior to network signaling and targeting to reduce the data used for knowledge extraction. CIP could be considered as an integral part of the afore described profiling schemes, i.e. COmpAsS and CEPE. The module comprises aggregating and compressing mobile network-related context information per unique identifier, such as the end device’s International Mobile Subscriber Identity (IMSI), as well as techniques related to identifying and discarding user profile-redundant or unnecessary context data, before any transmission to CEPE. CIP gains are illustrated via a detailed analytical approach, guided by well-established 5G use case requirements.As a final major contribution of this thesis, a comprehensive analysis takes place with regard to the CEPE-COmpAsS interworking, in the context of the forthcoming 5G network architecture and by mapping them with the latest 5G network components –as these were introduced in the latest 3GPP work-. The work in this section shows how the proposed framework can be instantiated as part of the 5G network components and functions introduced in SDN-enabled environments, such as the Network Slicing approach, the Network Data Analytics and the Network Slice Selection Functions, towards further optimising the distribution and management of the available infrastructure and network resources among the UEs, as well as the Access Traffic Steering, Switching and Splitting (ATSSS), responsible for managing the UE data flows and mapping each single UE flow with the optimal available access technology.Two supplementary studies are finally included in this dissertation: a preliminary analysis on traffic engineering policies based on user profiling realised by CEPE, as well as a 5G use case related to the Internet of Things domain -and more specifically, Precision Farming-, aiming to highlight explicit requirements such as mission-critical machine type communication.
Degree
thesis:*- Grantor dc:publisher
- National and Kapodistrian University of Athens
- Year dc:date
- 2018
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Barmpounakis, Sokratis
- Μπαρμπουνάκης, Σωκράτης
Subjects
dc:subject × 20- Δίκτυα 5ης γενιάς
- Διαχείριση Πόρων Ασύρματου Δικτύου
- Επιλογή Ασύρματης Τεχνολογίας Πρόσβασης
- Προφίλ χρηστών
- Μεταπομπή/ Παράδοση Υπηρεσίας
- Πλαίσιο
- Δικτύωση Βασισμένη στο Λογισμικό
- Τεμαχισμός Δικτύου
- 5G
- Radio resource management
- Radio Access Technology selection
- User Profiling
- Handover
- Context
- Software defined networking (SDN)
- Network Slicing
- Φυσικές Επιστήμες
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Natural Sciences
- Computer and Information Sciences
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/43446