Network Slice Management in 5G

Introduction to Network Slice Management

• Focuses on management and orchestration of network slices within 5G, Fog, Edge, and Cloud environments
• Explores design/implementation of networks for simultaneous essential connectivity
• Notes the challenge of meeting performance needs of applications with a single network function set
• Highlights the 5G infrastructure public-private partnership (5G-PPP) and its identification of use cases
• Use cases include enhanced mobile broadband (eMBB), massive machine-type communications (mMTC), and ultra-reliable low-latency communication (uRLLC)
• eMBB needs high bandwidth to transmit data for video on demand
• mMTC typically uses many low throughput devices such as the Internet of Things(IoT)
• Cost-efficient solutions comprise slicing physical network into isolated logical networks
• Network slicing is virtualization that partitions a shared physical network to end-to-end logical networks
• Network slicing enables 5G networks where vertical service providers deploy applications flexibly
• Network-as-a-service (NaaS) allows service providers to build customized networking infrastructure

SDN and NFV for Network Slicing

• Software-defined networking (SDN) and network function virtualization (NFV) serve as building blocks for network slicing
• They allow for facilitating network programmability and virtualization
• SDN separates control and data planes, providing centralized network oversight for running control functions
• NFV transfers network functions from hardware to software applications, reducing cost and increasing network function elasticity
• Virtual network functions (VNFs) are chained to build communication services

Chapter Overview and 5G Background

• Chapter reviews current network slicing literature in 5G, edge/fog, and cloud computing
• It aims to identify challenges to achieving ultimate network slicing realization
• Starts with intro to 5G, edge/fog, and clouds and their interplay
• Outlines 5G vision for network slicing and identifies a framework for it
• Reviews network slicing research in cloud computing
• Focus is on SDN and NFV technologies
• Explores network slicing advancements in fog and edge cloud computing
• Identifies unresolved network slicing challenges within platforms and discusses gaps
• 5G is proposed as telecommunications standards beyond 4G
• 5G has 802.11ac IEEE wireless networking criterion and operates on millimeter wave bands
• 5G encapsulates extremely high frequency (EHF) from 30 to 300 gigahertz (GHz), which offers data capacity and low latency communication
• 5G is expected to be mature by 2020
• 5G's main intentions include enabling Gbps data rate and offering long-term communication
• 5G aims to improve energy usage and be more flexible, dynamic, and manageable

Cloud and Mobile Edge Computing

• Cloud computing is expected to be an inseparable part of 5G services
• Cloud computing is considered a successful computing paradigm for delivering on-demand services
• Cloud data centers adopted virtualization for efficient management of resources
• SDC utilizes cloud computing, system virtualization, SDN, and NFV
• It aims to enhance resource management in data centers
• Cloud is the foundation for cloud radio access network (CRAN)
• CRAN is a cellular framework that aims to meet growing end-user demand on 5G
• In CRAN base stations are split into radio/baseband, where the baseband part is placed in the cloud for different base stations
• Mobile Edge Computing (MEC) is considered an enabler of 5G
• In MEC, base stations and access points are equipped with edge servers to handle 5G-related issues
• MEC facilitates computationally enriched distributed RAN architecture upon LTE based networking
• Ongoing research targets real-time context awareness, computation offloading, energy efficiency, and multi-media caching for 5G networking

Edge, Fog Computing, and Network Slicing

• Edge and fog computing complement the remote cloud
• It meets service demands of a geographically distributed large number of IoT devices
• Edge computing uses embedded computation capabilities of IoT devices to process data
• It is suited for light computational tasks without probing the global internet unless needing remote cloud
• Executing latency sensitive IoT applications at remote cloud degrades QoS
• Fog computing offers infrastructure and software services through distributed fog nodes to execute IoT applications
• Traditional networking devices like routers, switches, set-top boxes, and proxy servers can act as fog nodes
• Mobile edge servers or cloudlets are regarded as fog nodes to conduct respective jobs in fog-enabled MCC and MEC
• Edge and fog computing are used interchangeably, but edge is also is considered as a subset of fog computing
• Integration of 5G has been discussed in terms of bandwidth management during computing instance migration
• Integration of 5G has also been discussed in terms of SDN-enabled IoT resource discovery
• Concept of fog radio access network (FRAN) is getting attention where fog resources create BBU pool for base stations
• Network slicing is a key feature of 5G network virtualization that can extend vision to data center and fog nodes
• Network slicing can be applied to shared data center network infrastructure and fog networks
• This provides an end-to-end logical network for applications by establishing virtualized environment

Network Slicing in 5G: Definition and Benefits

• Network slicing in 5G is sharing a physical network’s resources to multiple virtual networks
• Network slices are virtualized networks on top of a physical network
• Network slices meet the requirements of specific applications, services, use cases, or business models
• Each network slice operates independently, with own resources, topology, data traffic flow, management policies, and protocols
• End-to-end implementation supports coexistence of heterogeneous systems
• Network slicing paves customized connectivity among interconnected end-to-end devices
• It enhances network automation and leverages SDN and NFV
• Makes traditional networking architecture scalable and shares underlying infrastructure to multiple virtualized networks
• Network slicing reduces capital/operational expenses and ensures reliability
• It assists isolation and protection of data, control and management plane that enforce security

Network Slicing Challenges and Architecture

• 5G network slicing faces challenges like resource provisioning among multiple virtual networks
• Each network has different resource affinity that changes over time
• Mobility management/wireless resource virtualization can intensify network slicing problems
• End-to-end slice orchestration and management can complicate network slicing
• Recent research in 5G network slicing aims to address these challenges through efficient frameworks
• A generic framework consists of infrastructure, network function, and service layers
• The infrastructure layer defines the physical network architecture that expands from edge to remote cloud
• encapsulation of software-defined techniques that facilitate resource abstraction
• Policies in infrastructure layer deploy/manage the underlying infrastructure and it allocates resources to network slices
• Function and virtualization layer manages virtual resources/network function life cycle
• It facilitates optimal placement of slices to virtual resources and chaining of multiple slices to meet needs
• SDN, NFV, and virtualization techniques are technical aspects of this layer
• Explicitly manages functionality of core/local radio access network and can handle coarse/fine-grained network functions efficiently
• Service and application layer is composed of connected vehicles, virtual reality appliances, mobile devices, etc
• It represents utility expectations from networking infrastructure and functions with virtualized network functions mapped to physical resources
• Virtualized network functions ensure SLA for application/service
• Slicing Management and Orchestration (MANO) explicitly monitors/manages all the layers with three main tasks:
  • Create virtual network instances upon the physical network using infrastructure layer functionality
  • Map network functions to virtualized network instances to build a service chain
  • Maintain communication between service application and the network slicing framework

5G Network Slicing and Cloud-Native Architecture

• Logical framework of 5G network slicing is still developing
• Extension of the framework can handle future dynamics for standardization of 5G
• According to Huawei, Cloud-Native network architecture for 5G has four characteristics
• It provides cloud data center-based architecture and independent network slicing on infrastructure to support scenarios
• It uses Cloud-RAN to build radio access networks (RAN)
  • Cloud-RAN provides many connections and implements 5G required RAN function deployments
• It provides simpler core network architecture, on-demand configuration of network functions, unified database management, and component based functions
• It automatically implements network slicing service to reduce operating expenses

Network Slicing in Software-Defined Clouds

• Review details state-of-the-art work on network slice management happening in cloud computing literature
• Survey can help researcher apply advances and innovation mutually in 5G and clouds
• Virtualization is cornerstone of resource management/optimization in cloud data centers
• Many research proposals exist for VM placement and migration to improve server utilization and efficiency
• Focus is on network-aware VM/VNF management aligned with network slicing management for SDCs
• Proposed taxonomy diagram classifies existing works based on research objective and approach to address the problem

Network-Aware Management Techniques

• Taxonomy classifies existing works based on the objective of the research - Approach used to address the problem - Exploited optimization technique - Evaluation technique to validate the approach
• Covered is network slicing from three different perspectives
  • Network-aware VM management
  • VM migration
  • VNF management.
• Cziva et al. present orchestration framework to exploit time-based information to migrate VMs and minimize cost
• Wang et al. propose a VM placement mechanism reducing the number of hops between VMs, save energy, and balance load
• Remedy relies on SDN to monitor network state and estimate VM migration cost Detecting congested links and migrating VMs
• Jiang et al. worked to minimize network cost in real-time with online algorithm that optimizes VM placement and traffic routing
• VMPlanner optimizes VM placement/network routing, including VM grouping that consolidates VMs, placement in a rack, and traffic consolidation.

VM Migration and Functions

• Jin et al. studied joint host-network optimization problem
• The problem is formulated as an integer linear problem (ILP) that combines VM placement and routing
• Cui et al. explore joint policy-aware and network aware VM migration problems VM management reduces network wide communication costs
• Bari et al. proposed a method for finding efficient migration plans
• Monitors residual bandwidth available on the links between source and destination after performing each step
• Ghorbani et al. generated algorithms to generate an ordered list of VMs to migrate and forward flow changes
• Concentrate on imposing bandwidth guarantees on the links to ensure that link capacity is not violated during migrations
• Li et al. addressing the workload-aware migration problem and Selection methods for VMs, hosts and sequence for migration
• These studies focus on the migration order of a group of VMs while taking into account network cost.
• Xu et al. proposed an interference-aware VM liver migration called Aware that minimizes migrations and co-location

Virtual Network Function Management

• NFV is an emerging paradigm where network functions such as firewalls/NAT and virtual private networks (VPNs) are divided up into building blocks called VNFs
• VNFs are chained to build service function chains (SFC) to deliver required network functionality
• Han et al. present survey of key challenges/technical requirements for NFV, presenting architectural framework for it
• Focus is on the efficient instantiation, placement and migration of VNFs and network performance
• VNFs are a model used for efficient placement of VNFs, as proposed by Moens and Turck
  • They propose a NFV burst scenario in a hybrid scenario in which baseline need for network function comes from physical resources, while extra load is handled by virtual ones
• Cloud4NFV is a platform by the European Telecommunications Standards Institute (ETSI) to build network function as a service using a cloud platform with a VNF Orchestrator exposing RESTful APIs that allow VNF deployment
• Supports management of virtual infrastructure in background with OpenStack
• vConductor is another NFV management system proposed by Shen et al. for end-to-end virtual network services that has gui's and supports management of VNFs and existing physicals
• Yoshida et al. proposed vConductor (building NFV using VMs) in the presence of conflicting objectives
• A service chain is a series of VMs hosting VNFs in an order, with flow going through them to get desired network functionality
• Tabular VM migration (TVM) proposed by Aim aiming at reducing the number of hops in service chains of network functions (cloud data centers) to satisfy SLAs with SLA driver variable width
• SOVwin is a heuristic in addressing that problem.

NFV Projects

• EU-funded T-NOVA project, realizing the NFaaS concept, has designed and implemented integrated management and orchestration platforms (auto provisioning/management)
• UNIFY is another EU-funded project aimed at supporting automated, dynamic service creation:
  • Based on fine-granular SFC model, SDN, and cloud virtualization techniques
• Fog computing: attempts to address quality of service (QoS) requirements of applications which require real-time applications

Edge and Fog Computing

• Fog computing serves as a middle layer between edge and core clouds
• Fog serves apps close to the data source and the core clouds offer massive duty
• SDN and NFV play key roles in prospective solutions facilitating the effective management of netowrk services
• Despite synergy + kinship - less research exists in integrating fog/edge computing + SDN/NFV
• Lingen et al: define model-driven, service-centric architecture to integrate NFV, fog and SD MEC that aligns uniform management of services from cloud to Lingen et al
• Two-layer abstraction model with IoT -specific modules to intergrate cloud and Truong et al - earliest to propose an SDN - based arcitecture to support fog computing
• Proposed architecgure centraizes network by ultilxing the controller