ABSTRACT
The growth of the telecommunication industry is fast paced with ground-breaking engineering achievements. Despite this, portable mobile handheld devices have very low computational, storage and energy carrying capacity occasioned by the needs to satisfy portability, very small form factor, ergonomics, style and trends. Proposals such as cloudlets, cyber foraging, mobile cloud computing (MCC), and more recently but most applicable, multi-access edge computing (MEC) have been proffered. New and emerging use cases, especially the deployments of 5G will bring up a lot of latency- sensitive and resource-intensive applications. To address these challenges, this work introduced the use of secure containerization for MEC applications and location of MEC host at the 5G centralized unit within the radio access network (RAN) aiding offloading of computational, storage and analytics requirements close to UE at the fringe of the network where the data are being generated and results being applied. The major contribution of this thesis is the use of secured containerization technology to replace virtual machines, making it possible to use containers for MEC applications, reducing application overhead while satisfying the isolation of MEC infrastructure as required by the European Technology Standard Institute (ETSI) to ensure MEC application security. 5G end-to-end transport specifications were evaluated for the vantage location of MEC server within the radio access network and achieved theoretical values between 4.1ms and 14.1ms end-to-end latency. These figures satisfy requirements of VR/AR (7-12ms); tactile Internet (<10ms); Vehicle-to-Vehicle (< 10ms); Manufacturing & Robotic Control/Safety Systems (1-10ms). The results confirmed that edge computing has lower user plane latency figures and reduced backhaul traffic with lower application failure rate. Secured containerised Multi-access computing infrastructures have many advantages of mobile cloud computing for mobile wireless device computational power and energy carrying capacity deficiencies to cater asymmetric UE applications. Applications hosted within the RAN have better support for new and emerging application requirements in terms of high amount of computational, storage and analytics capabilities while at low latency figures. Edge deployments will reduce the pressure on network operators backhaul link saving end-to-end ecosystem from collapse due to heavy backhaul traffic that might result from billions of 5G UEs. No matter how fast 5G network will be, MEC will ensure the need not to transport huge data for processing in the cloud and returning the results to the UE. This will enhance privacy and security while also conserving bandwidth
CHAPTER ONE
1.0 INTRODUCTION
1.1 Background to the Study
There are several constraints on mobile devices as well as other portable 5G user equipment (UE) devices. Computational resources, memory limitation, storage, network and energy carrying capacity are some of the constraints of cellular mobile communication UEs and these have a significant effect on the type of application software available and for how long a battery can hold a charge to support such applications. The major constraints include computational power, charge holding time, storage and memory limitations, especially for complex processes (Taleb et al., 2017). There are latency-critical and resource-intensive services which are needed to be supported by 5G, these include: telepresence, robotics, factory automation, and intelligent transportation systems, Virtual Reality (VR), Augmented Reality (AR), medicals, smart grid, serious gaming, education and culture (Parvez et al., 2018). More so, high definition images and multi-feed super high definition quality live video streaming demands for mobile users are constantly being escalated over the recent decade. 5G is projected to provide services that will support communication, computing, control and content delivery for high-intensity network traffic (Mao et al., 2017).
The deployment of 5G cellular mobile telecommunication standard will herald explosive evolution of Information and Communication Technology (ICT) innovations for mobile devices, machine-to-machine, Internet of Things (IoT), emerging vehicle technologies – V2V and V2X, etc. There will be an enormous increase in the number of mobile devices, expectedly about 50 billion devices, but this number will be completely dwarfed by the exponential growth in the volume of data generated by resource-intensive and feature-rich multimedia applications. These will create hype for mobile data traffic and compute requirements (Skarpness, 2017).
To resolve challenges posed by these constraints, the computational requirements of mobile applications were offloaded to be processed on tethered external infrastructures with adequate resources. These external infrastructures are usually commercial off-the- shelf or customized standardized IT infrastructures configured to process and return results for applications. Different interventions have been proposed, including cyber foraging, cloudlet, multi-access edge computing (MEC) and mobile cloud computing (MCC) (Mach and Becvar, 2017).
A cloudlet is a resource-rich computer or cluster of computers that is well connected to the Internet, trusted and is constantly available for use by untethered mobile devices within close proximity (Satyanarayanan et al., 2009). Cyber foraging is considered as dynamical augmentations of computing resources of mobile devices and opportunistically exploiting computing of tethered infrastructure in the nearby environment (Satyanarayanan, 2001). It is the capability of infrastructure to seamlessly undertake migration of computation from one node to another (Patil et al., 2016). Cloud computing (CC) is the abstraction of computing resources e.g. processor, RAM, storage and network services from separate hardware units while presenting as a pool of reusable on-demand shared computing infrastructure. This can be provisioned rapidly and released programmatically or manually involving minimal management effort while creating cost benefits and flexibility. The rate of cloud adoption is very high, majorly making it an economically viable option. MCC is the integration of CC to serve cloud-based web apps over the Internet for smartphones, tablets, and other portable devices. Cloud computing in mobile cellular networks, like every other technology, though a solution, has come with its fair share of challenges as cellular mobile communication technologies mature from
2.0
1G, 2G, 3G to 4G and looks forward to 5G. Ordinarily, cloud computing should provide enough resources for offloading of computational demands (Mach and Becvar, 2017). Cloud computing is predominantly application programming interface (API) driven and economically viable. Cloud infrastructure can be shared across many end users, developers, mobile network operators and corporations spanning over widely separated geographical locations, allowing the reduced cost of services compared to traditional legacy infrastructures. Despite all the potentials of cloud computing, it has been unable to fulfil mobile application end-to-end latency requirements due to long response times, due, in turn, to the centralized cloud architecture model resulting into high signal propagation delay, affecting the end-user quality of experience (QoE) (Taleb et al., 2017). Other concerns presented by the use of cloud computing include security and privacy, addressing, interoperability, bandwidth (Díaz et al., 2016), as well as government policy (National Information Technology Development Agency (NITDA), 2019); (Okafor, 2017).
The advent of the so much anticipated 5G technologies emerging mobile applications such as Augmented Reality (AR), Virtual Reality (VR) (Alsafi and Westphal, 2016), face/voice detection and identification for surveillance, authentication and access control, connected autonomous vehicles (CAV), intelligent transportation systems (ITS) and highway traffic management systems, ultra-high-definition multi-feed live streaming. These are anticipated to be among the high resource demanding applications over wireless cellular networks. In particular, the newly emerging mobile Augmented Reality and Virtual Reality (AR/VR) applications are anticipated to be among the most demanding applications over cellular wireless networks (Erol-Kantarci and Sukhmani, 2018).
Field devices such as traffic signals, roadside sensors, face detection and identification, surveillance networks, location services, Intelligent Transportation Systems (ITS) and 3 highway traffic management systems. are gradually being connected to central monitoring systems for better traffic management. MEC seeks to address issues and challenges surrounding billions of field devices generating gigabits of low latency- sensitive data that require split seconds between data generation, data processing and eventual results being applied to certain control mechanisms. Advantages of edge computing include:
A. Access to real-time radio network information that creates opportunities that can be leveraged by applications (Shahzadi et al., 2017) giving rise to location-based services opportunities like location-aware advertising, asset tracking, connected autonomous vehicle, AR, ITS and highways traffic management systems.
B. Edge computing will improve Quality of User Experience (QoE) by leveraging on reduced latency and high throughput available when application service logics are computed on the edge servers within the RAN
C. Reducing data security breach and enhancing privacy by reducing the level of exposure through manipulation of data close to the source rather than transmitting via numerous routes to the cloud.
D. Edge computing creates a new and emerging market value chain in mobile networks thereby opening the network to third parties (Patel et al., 2014), who can develop and quickly create innovative applications, benefiting all parties (Huang et al., 2017).
1.2 Statement of the Research Problem
The motivation for this research work stems from the claim by Satyanarayanan et al (2009) that “resource poverty is a fundamental constraint that severely limits the class of applications that can run on mobile devices.” It was also argued by Satyanarayanan et al. (2009) that “at any given cost and level of technology, considerations of weight, size, battery life, ergonomics, and heat dissipation exert a severe penalty on computational resources such as processor speed, memory size, and disk capacity.” There are challenges of unacceptable latency figures in 4G deployments but in 5G which has higher traffic, it is feared that there might be an increase in latency. These have been perennial challenges to mobile communication UEs but the deployment of 5G is expected to further exacerbate the problems with the rise of new and emerging feature-rich mobile applications generating high-intensity network traffic. Besides the challenges posed to mobile UEs, this scenario will be exerting unprecedented pressure on backhaul and fronthaul networks.
Researches have been carried out to augment for computational resources as well as energy-carrying capacity constraints in the mobile wireless devices, but the shortcomings of the earlier proposed solutions include:
A. Unstable Infrastructure:
The inadequate or total loss of stable augmenting infrastructure has been the bane of cyber foraging techniques to resolve the challenges of resource constraints in portable mobile UEs (Gordon et al., 2012); (Qing et al., 2013); (Dean and Ghemawat, 2004); (Huang et al., 2010); (ETSI, 2017). This usually creates a situation of low bandwidth that results in poor application behaviour. Whereas MEC is compatible with orchestration, software configuration management, IaC and VCS, It is also deployable on standard off-the-shelf or customized IT infrastructure making it possible to have vertical and horizontal infrastructure scalability, the mechanisms required to achieve stable infrastructure.
B. Platform-specific:
Mobile Assistance Using Infrastructure (MAUI) in Cuervo et al. (2010), Code Offload by Migrating Execution Transparently (COMET) (Gordon et al., 2012) and Misco – a MapReduce framework for mobile systems in Dou et al. (2010) relied on Microsoft .Net Framework; the now discontinued, Android Dalvik Virtual Machine and Python respectively limiting the type of mobile applications making use of the solutions. MEC can support computer programming language or frameworks consistent with standard or customized IT platforms that require computational, memory, storage and power resources (ETSI, 2019).
C. High mobility interruption time or zero mobility of augmenting infrastructure.
High mobility interruption time of 30 milliseconds in 4G LTE whereas cloudlets (Satyanarayanan et al., 2009) and MAUI in Cuervo et al. (2010) have zero support for mobility interruption. User mobility is not encouraged by these augmenting solutions. MEC on 5G will deliver mobility interruption time of zero millisecond (ITU, 2015); (Taleb et al., 2017).
D. High user plane (UP) and control plane (CP) latency:
High data communication delay between the UE and Internet/cloud for cloudlets and mobile cloud computing-based strategies in Zhang et al. (2010); Chun et al. (2011); Kosta et al. (2012); (Gordon et al., 2012). The total UP latency for MEC deployment in 4G Long Time Evolution Evolved Packet Core (LTE/EPC) is estimated, with reference to Parvez et al. (2018), as the summation of packet transmission delay between the UE via LTE, EPC to MEC host connected across the Steering GPRS interface (SGi) for services sourced at MEC. Average container boot time is 1.87 seconds compared to Virtual Machines (VMs) average boot time of 94.90 seconds (Zhang et al., 2018), containerized MEC can offer lower application control plane latency.
E. Inconsistencies between production and staging MEC environments: Containerized applications have the advantage of portability, running in the same environment during testing, staging and production deployments.
The issue of resource poverty has been a perennial challenge to the telecommunication industry. The scale of 5G will be enormous, both in terms of devices and data but the number of devices will be dwarfed by the volume of data will be generated (Skarpness, 2017). The huge amount of data, obviously, will require analytics as well as transport. There are already unacceptable latency figures in 4G LTE/EPC deployments but in 5G which has higher traffic, it is feared that there might be an increase in the latency.
1.3 Aim and Objectives of the study
The aim of this work is to demonstrate use cases for MEC deployments in 5G Networks by developing models of MEC infrastructure deployments.
The research objectives are to:
i. develop low latency models for MEC deployment for 5G networks through the evaluation of both 3GPP and non-3GPP components of 5G networks transport specifications.
ii. develop and deploy secure resource-intensive containerized mobile web application testbed implemented in Kata Containers (Kata Containers, 2019) for both edge and remote cloud servers.
iii. performance evaluation of the developed MEC models and determination of the best fit for 5G applications.
1.4 Justification of the study
The drawbacks of previous research works were explored. These include limitations posed by pre 5G wireless technologies on the maximization in meeting latency requirements of new and emerging applications consistent with International Telecommunication Union (ITU) specifications for 5G eMBB and uRLLC use cases.
1.5 Scope of the study
The research focused on only the Multi-access Edge Application part of Mobile Edge Host within the Multi-access Edge (ME) Host Level of ETSI MEC framework (ETSI, 2019a). This research work does not include any work on ME systems-level management, ME host-level management or network-level entities of MEC. This work also proposed the location of MEC host close to 5G centralized unit (CU), connected directly to packet data convergence protocol (PDCP). Disposable infrastructures required for mobile applications computation, storage and analytics deployed at the MEC are available as machine-readable definition files downloadable from synchronized but distributed repository with tracking and coordination of files modifications using Git.
In this work, the estimated user plane latency values were benchmarked with values of known low latency application use case requirements (Sutton, 2018b):
A. Virtual Reality & Augmented Reality: 7-12ms
B. Tactile Internet (Remote Surgery, Remote Diagnosis, Remote Sales): < 10ms
C. Vehicle-to-Vehicle (Co-operative Driving, Platooning, Collision Avoidance): < 10ms
D. Manufacturing & Robotic Control / Safety Systems: 1-10ms
Models for MEC based on infrastructure as code (IaC), containerization and version control system (VCS) deployment scenario for eMBB and uRLLC use cases were proposed for end-to-end 5G network, variously and concurrently serving multiple asymmetric mobile user computational requirements by leveraging on IaC, containerization and VCS to enable orchestrated provisioning, patching, freezing, caching, resuming and termination of infrastructure instances. This will enable the MEC server to continue to programmatically serve different mobile applications as at when required and free up resources when applications are not being served.
1.6 Thesis Outline
This thesis is composed of five chapters. In Chapter One, the work was introduced, motivation for the study of Container-Based Multi-access Edge Computing for 5g Networks, the aim and objectives were enumerated. Definition of terms and keywords, the research fundamentals, including the statement of the research problem were explained in this chapter.
The remaining part of this document is organized as follows: Chapter Two provides review of past research efforts to ameliorate the resources poverty challenges inherent in portable handheld mobile cellular user equipment. In Chapter Three – Methodology, both 3GPP and non-3GPP components of 5G networks transport specifications were evaluated. This chapter also contains the detail and description of the techniques used in the experiments. Results of the technical evaluation of 5G transport specifications are specified and discussed in Chapter Four. Results from the experiment comparing the behaviour of applications deployed at the MEC relative to mobile cloud computing (MCC) deployments were equally presented in Chapter Four. Chapter Five contains the conclusions arrived as a result of this research work. Included in Chapter 5 are the research limitations and future works, contributions and recommendations.
This material content is developed to serve as a GUIDE for students to conduct academic research
A1Project Hub Support Team Are Always (24/7) Online To Help You With Your Project
Chat Us on WhatsApp » 09063590000
DO YOU NEED CLARIFICATION? CALL OUR HELP DESK:
09063590000 (Country Code: +234)
YOU CAN REACH OUR SUPPORT TEAM VIA MAIL: [email protected]
09063590000 (Country Code: +234)
