Back to search

University of Thessaly (UTH)

Design and implementation of resource allocation algorithms in 5G heterogeneous wireless networks

Abstract

dc:description

Next generation (5G) and beyond mobile networks will highly rely on distributed heterogeneous infrastructure, which is foreseen to create a versatile resource ecosystem. Heterogeneous Networks (HetNets) have already been proposed as a solution for the continuously deteriorating congestion problem of cellular infrastructures. Current mobile devices are able to utilize the HetNets since they are equipped with multiple radio access interfaces. Still the network selection problem in a HetNet is an open research issue. At the same time, 5G redefines the network architecture, through the introduction of base station disaggregation and the softwarization of the Radio Access Network (RAN). These features stem from the Cloud-RAN concept, where a part of the base station is running as a software in the Cloud and the rest on the hardware connected to the RF transceiver. These Cloud-RANs might aggregate multiple heterogeneous links on the access, allowing users to access the network with different technologies. In such network deployments, efficient traffic allocation and management is of prominent priority. Another important advancement of 5G, is the minimization of latency for accessing services through the application of the Multi-access Edge Computing (MEC). This functionality brings the services at the edge of the network, supporting several applications that exchange time-critical data over the network (e.g. e-Health, Industry 4.0, AR/VR, etc.). In this thesis, we design, study and propose innovative solutions for the resource allocation in 5G Heterogeneous Wireless Networks (with Base Stations either single unit or disaggregated) and with MEC capabilities. The fundamental questions that we try to answer are the following: 1) How should a UE choose the best Radio Access Technology (RAT) in a heterogeneous network and when should it change to a different RAT? 2) How should an operator of such a network charge each of the available RATs? 3) Given a heterogeneous Cloud-RAN, how can we serve end-users with low latency? 4) How should resources from a MEC enabled Cloud-RAN network be allocated to different Service Providers? 5) How would an operator be charged for moving services closer to the edge for providing lower latency access to the users?Initially we begin with the ultra-dense deployments of heterogeneous wireless networks. Network densification and integration of heterogeneous technologies have been proposed as a solution to the constantly rising number of devices connected to the Internet. Heterogeneity is considered as a key feature of the 5G networks as it enables the mobile service providers to offload their cellular networks, thus mitigating the existing congestion and providing high quality connectivity to mobile users. Nevertheless, such environments add up to the complexity of the traffic allocation problem. To this aim, we design and study a heterogeneous network access scheme based on the Paris Metro Pricing (PMP), considering three different access technologies (3G, 4G, WiFi). We introduce a dynamic pricing scheme for each available RAT of the heterogeneous network which takes into consideration the current available capacity of each RAT. We investigate its performance in terms of average throughput and acceptance capability of incoming users for several mobility scenarios, employing simulations and real-testbed experiments. As our next step, we study the use case where the end-users are multi-homed, meaning that they feature multiple network interfaces and can be concurrently connected to different RANs. Moreover we consider that the end-users are running applications with different network demands (low latency, throughput, etc.). In such setups, the technology allocation problem becomes even more complicated because the perceived Quality of Experience (QoE) highly depends on the application that is running on the end-user. To this end, we propose a scheme that tries to efficiently map each application to the available technologies towards maximizing the end-user's utility function. Through the evaluation of our solution, we pinpoint the significance that the end-users' order has to the system during its updating phase. For this reason, we propose and examine three different ordering policies (based on data rate demands, spectrum efficiency, and sensitivity). Our results illustrate how the ordering policies may affect the cost incurred at each UE as well as the utilization of the available RATs. Following this, we focus on the fairness of the UEs' ordering policies that we have introduced in our resource allocation scheme for 5G Heterogeneous Wireless Networks. Towards this direction, we introduce an optimal ordering policy as a reference. This policy leverages the weighted proportionality fairness concept. We compare it against the other proposed ordering policies taking the Kendall tau correlation for each one of the policies. Based on our findings, we design and implement a dynamic policy selection scheme, which selects among the available ordering policies the one which is closest to the reference policy every time that the system is updating. We further investigate the pricing of our scheme; initially, only the availability of access technologies’ bandwidth was taken into consideration for our pricing model but through the evaluation of our solution, we decided to extend it in order to also include the rate demands of the applications running on the end-user side. For the evaluation of our solution, we consider 4 different RATs (3G, WiFi, 4G, WiGig/5G-NR) for our 5G HetNet. Both simulations and testbed experimentation are employed and enable us to assess the bandwidth allocation efficiency, the variation of access prices, the policy fairness and the induced cost to end-users. The next part of our work is related to the minimization of service access latency. ETSI has proposed Multi-access Edge Computing (MEC) as the means to minimize the access latency by bringing the services closer to the wireless network edge. 5G-New Radio (NR) specifications define the disaggregation of the base stations based on the 3GPP Option-2 split which creates two entities: the Central Unit (CU) and the Distributed Unit (DU) which provides the actual access network. The CU is able to manage multiple DUs which can be of different access technologies (5G-NR, 4G LTE or even WiFi). Towards this direction, we study the placement of MEC services in a disaggregated network (Cloud-RAN) with multiple wireless technologies. We investigate two different deployments of the services, either on the fronthaul collocated with the DU or the backhaul collocated with the Core Network and for different access technologies (LTE, WiFi). Our testbed experiments denote that placing the service close to the base station reduces the service access latency even with legacy technologies (LTE, WiFi). Furthermore, we investigate how the UE's Quality of Experience (QoE) is affected from the service placement in conjunction with the wireless technology that the UE will be served. Finally, we study the resource allocation of MEC resources in 5G Heterogeneous Cloud-RANs. In such setups, multi-homed users are able to get access over multiple wireless links to services located just after the DU components of the cellular network, illustrating reduced network latency compared to conventional MEC deployments. Nevertheless, the technology through which each user may be served plays an important role in the overall perceived latency and Quality of Experience (QoE) of the end-user. Moreover, the locations for placement of the hosted services and wireless technologies used to forward data to the end-users can be exploited as differentiation parameters for charging application providers for hosting their services on the MEC platform. Therefore, we design and implement a system model for resource allocation across the different supported MEC combinations on a Cloud-RAN with LTE and WiFi technologies. Our testbed experiments highlight that utilizing multiple RATs at the same time and with the appropriate placement of the MEC service, the average latency can be kept below a threshold while meeting the service's requirements.

Degree

thesis:*
Grantor dc:publisher
University of Thessaly (UTH)
Year dc:date
2021

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Passas, Virgilios
  • Πασσάς, Βιργίλιος

Subjects

dc:subject × 15

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/49036

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Passas, Virgilios; Πασσάς, Βιργίλιος. Design and implementation of resource allocation algorithms in 5G heterogeneous wireless networks. University of Thessaly (UTH), 2021. http://hdl.handle.net/10442/hedi/49036