Reliable Slicing in 5G Networks

Luca Valcarenghi, Alessio Giorgetti, Barbara Martini, Koteswararao

Kondepu, Molka Gharbaoui, Piero Castoldi

Abstract In 5G, the term slicing refers, in general, to the possibility for
different customers (usually called tenant) to share the same physical network.
Thanks to the softwarization of networks according to the Network Function
Virtualization (NFV) concept and the programmability of network
connectivity through Software Defined Networking (SDN), new network and
service capabilities can be envisioned by integrating networking, computing
and storage resources while serving a multitude of tenants. Each tenant is
assigned a logical network that can satisfy its requirements. Survivability is
one of the most important requirements especially for vertical applications
requesting Ultra Reliable Low Latency Communications (URLLC). In this
chapter the concept of slice is introduced and the some use cases for providing
reliability in a slice.

1 5G Slicing
With the advent of NFV and SDN a novel network scenario is envisioned
enabled by network deployments into the cloud also extended to the network
edge and by programmability of network connectivity through network
controllers. This trend known as softwarization is enabling new unique
network and service capabilities by integrating networking, computing and
storage resources into one programmable and unified infrastructure while
serving a multitude of distributed smart devices and applications (e.g., robots,
drones, smart vehicles). As result, current communications network scenario
is moving from having a separate network for each application (e.g., fixed
telephone network, mobile telephone networks, Internet access) to a single
network shared by different applications or verticals. Network Slicing is a key
 2 Valcarenghi, Giorgetti, Martini, Kondepu, Gharbaoui, Castoldi

feature of the 5G System that allows Operators to flexibly structure the

network resources to match the services offered to subscribers, third-party
customers, including the roaming scenario. The concept of slicing emerged as
a way of setting up several logical networks for different verticals on the same
physical network. Each vertical is then assigned to the logical network that
guarantees the required QoS. Such setup potentially allows communication
providers to save capital and operating expenditures (CAPEX and OPEX).
However, as for any shared medium, guaranteeing the required QoS to
network slices sharing the same physical network is not a trivial task and
remains an open issue. In particular, slice control and management planes
shall be designed for slice provisioning and dynamic reconfiguration and the
data plane shall guarantee each slice requirements (e.g., QoS requirements,
slice isolation, etc.).

This chapter overviews how the concept of slice is defined in different

Standard Developing Organizations (SDOs) and research projects.

1.1 The Concept of Slice in SDOs and Research Project

Several SDOs are focusing on the network slicing concept 1. The Next
Generation Mobile Networks (NGMN) alliance defines a Network Slice
Instance (NSI) as “… a set of network functions, and resources to run these
network functions, forming a complete instantiated logical network to meet
certain network characteristics required by the Service Instance(s)” 2. In 2
the network slicing concept consists of three layers depicted in Figure 1:
Service Instance Layer, Network Slice Instance Layer, and Resource layer.
The Service Instance Layer represents the services (i.e., end-user or business
services) which must be supported. The Network Slice Instance Layer
provides the network slice instances with specific network characteristics that
are required by the related Service Instances (e.g., Enhanced MBB, M2M,
Enterprise and Industry). The Resource Layer provides the physical or virtual
resources for slice deployment.
 Reliable Slicing in 5G Networks 3

Figure 1 Slice concept in 2

3GPP in TR 28.801 3 defines, in accordance with NGMN, a network slice

instance (NSI) as “… a set of network functions and the resources for these
network functions which are arranged and configured, forming a complete
logical network to meet certain network characteristics…”. In addition, it
defines the following phases of a network slice lifecycle: preparation phase,
instantiation, configuration and activation phase, run-time phase,
decommissioning phase. Moreover, it introduces three management functions
to manage the NSIs to support communication services: Communication
Service Management Function (CSMF), responsible for translating the
communication service related requirements to network slice related
requirements; the Network Slice Management Function (NSMF), responsible
for management and orchestration of NSI; and the Network Slice Subnet
Management Function (NSSMF), responsible for management and
orchestration of a network slice subnet instance (NSSI). Finally, it defines the
different roles of the actors (e.g., costumers, providers, operators, etc.)
involved in slice provisioning as depicted in Figure 2. However, TR 28.801
does not specify how to implement such functions and their relationship with
respect to the ETSI NFV architectural framework.
 4 Valcarenghi, Giorgetti, Martini, Kondepu, Gharbaoui, Castoldi

Figure 2 High level function of roles in 3

ETSI NFV EVE012 4 establishes the correspondence between a network slice

(3GPP) and a network service (ETSI NFV). There, ETSI describes that an
NFV Network Service (NFV-NS) can be regarded as a resource-centric view
of a network slice, for the cases where a NSI would contain at least one
virtualized network function. Moreover, ETSI NFV EVE012 proposes that
3GPP slice management functions interact with ETSI NFV Architecture
through the Os-Ma-Nfvo reference point as depicted in Figure 3. However,
in 4 it is stated that “…3GPP slice-related management functions are still
under definition in 3GPP SA5 and future updates might require further
analysis about the interaction between 3GPP slicing related management
functions and NFV-MANO…”.
 Reliable Slicing in 5G Networks 5

Figure 3 Interaction between slice management functions (3GPP) and

ETSI NFV Architecture from 4

Additional definitions of slices have been proposed by the following

organizations:
• The Internet Engineering Task Force (IETF) in 5 6 7
• The Broadband Forum (BBF) in 8
• The Optical Networking Forum (ONF) in 9
• ITU-T through the Focus Group on IMT (International Mobile
Telecommunication)-2020 (FG IMT-2020) in 10.
• Metro Ethernet Forum (MEF) in [11][12]

Within the research community, the 5G-Transformer project 14 envisions

three functional layers for providing verticals with slices: a Vertical Slicer as
the logical entry point for verticals to support the creation of their respective
transport slices in a short time-scale (in the order of minutes), a Service
Orchestrator to orchestrate the federation of transport networking and
computing resources from multiple domains and manage their allocation to
slices, and a Mobile Transport and Computing Platform (5GT-MTP), that
provides and manages the virtual and physical IT and network resources on
 6 Valcarenghi, Giorgetti, Martini, Kondepu, Gharbaoui, Castoldi

which slices are deployed. Such architecture implementation in under

development.

The 5G!Pagoda research project in 15 overviews most of the ongoing network

slicing-related activities. In addition, it proposes a functional architecture of
the slicing system and it delineates key implementation elements. The
functional architecture of the slicing system for the single domain is based on
a Domain Specific Slice Orchestrator (DSSO), a Slice Operations Support
(SOS), a Slice Management Plane (SMP), a Slice Software Layer (SSL), a
Slice Resource Layer (SRL), a Virtual Computing/Storage/Connectivity
Infrastructure Layer (VCSCI), and a Physical
Computing/Storage/Connectivity Infrastructure Layer (PCSCI).

Within the general architecture proposed by the research project SONATA

16, a “Slice Management Functional Block” is defined and integrated in the
NFV Management and Orchestration (NFV MANO) functional block,
detailed in the ETSI NFV architectural framework. In addition, the
implementation of such functions has been developed within the project.

2 Reliability in 5G Slices

The envisioned 5G network architecture, including the Next Generation Core

(NG Core, i.e., the new Evolved Packet Core --- EPC --- for 5G) and the New
Radio Access Network (New RAN), will be heavily based on virtual network
functions (VNFs) [17]. Network function virtualization (NFV) enables an
easy introduction of new network services by adding dynamic
programmability to network devices (e.g., as routers, switches, and
applications servers) that, in turn, empowers fast, flexible, and dynamic
deployment of new network and management services. Moreover, network
function virtualization also enables network slicing by providing multiple
instances of the same network function. In this context, the dynamic service
chaining allows the delivery of a new breed of applications (e.g., cloud
robotics, smart cities) by dynamically selecting and composing computational
and network services deployed as virtual functions (VFs) in distributed micro-
clouds located at the network Edge closer to the users [25]. The exploitation
of network function virtualization is foreseen also in the NG core [18] and the
 Reliable Slicing in 5G Networks 7

New RAN technology [19]. In the NG Core, the different network functions
(e.g., Access and Mobility Function (AMF), Session Management Function
(SMF), Policy Control Function (PCF), Application Function (AF),
Authentication Server Function (AUSF), User Plane Function (UPF), and
User Data Management (UDM)) can be virtualized, as it has been proposed
for LTE-A [18], and placed in different virtual machines (VMs) or run as a
single bundle in one VM.

As specified in [19] “…Service continuity is not only a customer expectation,

but often a regulatory requirement …”. Thus, in a scenario where network
functions are virtualized, both hardware and software failures assume the
same importance, and their reliability shall be guaranteed. Similarly,
reliability at service chain level is important to assure proper service
availability features to application service platforms deployed by verticals
[25][26].

2.1 Virtualised EPC reliability

In the technical specification document (TS) 23.007 [20], 3GPP specified

different failure detection and recovery mechanisms for EPC components,
including detection of path failure with the help of Echo Request/Echo
Response timer messages. Moreover, approaches for recovering failures in a
scenario where a mobile network function is virtualised can stem from
schemes already proposed for grid and cloud networking [21]. Furthermore,
scalable architectures for reliability management are being defined by ETSI
NFV [22] and implemented in current open source orchestration frameworks
such as Openstack [23]. However, the performance of resilience schemes
based on the aforementioned approaches once applied to 5G mobile network
have not been fully evaluated so far.

This section demonstrates the capability of recovering vEPC failures by

means of a vEPC in “hot backup”. Both working vEPC and backup vEPC are
deployed in multiple Network Function Virtual Infrastructure Points of
Presence (NFVI-PoPs) made available by the federated testbeds belonging to
the SoftFIRE project [24]. The demo is designed to evaluate the Service
Recovery Time (SRT), that is the time required to regain user equipment (UE)
 8 Valcarenghi, Giorgetti, Martini, Kondepu, Gharbaoui, Castoldi

connectivity, when the proposed resilient scheme is deployed in different

NFVI-PoPs.

The considered scenario and the proposed resilience scheme are depicted in
Fig. 4 and Fig. 5 by referring to functional elements of the Long Term
Evolution-Advanced (LTE-A) architecture.

The proposed resilience scheme considers a scenario where the vEPC fails
(e.g., a virtual machine where the vEPC runs crashes). Fig. 4 shows the two
considered vEPC resilience schemes based on vEPCs hot backup deployed in
federated NFVI-PoPs. The one on the left features two co-located vEPCs (i.e.,
vEPCa and vEPCb deployed in Surrey 5GIC testbed) while the one on the
right features a remote hot backup vEPC (i.e., vEPCr) deployed in a different
compute resource available in another testbed (i.e., Fokus ). In the latter case
two testbeds will be contemporarily utilized to implement the resilience
scheme. In the Surrey 5GIC testbed two different VNF functions (vOAISIM
and vEPC) will be implemented by exploiting open source mobile platforms
(i.e., OpenAirInterface-OAI). Here, vOAISIM VNF provides emulation of
virtual user equipment (vUE) and evolved NodeB (eNB) while vEPC will be
used to emulate the core network.

Fig. 5 shows the considered scenario and lifecycle event when vEPC VNF
fails. Here, when VNFs are deployed, vOAISIM connects with vEPCa, and
Zabbix server start monitors the VNFs that are associated corresponding
Zabbix agent. Note that each vEPC VNF and vOAISIM VNF deployement
contain also Zabbix agent. If the Zabbix server detects an anomaly activity in
vEPCa (e.g., overload) or does not receive any status report from vEPCa (i.e.,
vEPCa crashed) for a pre-defined period of time (i.e., time to trigger the
activity), the Zabbix server check the status of the hot backup vEPC to initiate
a recovery procedure. The receovery procedure consist in reconfiguring
vOAISIM to connect to the hot backup vEPCr. Upon reconfiguration
vOAISIM is able to communicate hot backup vEPCr. Similarly, the
experiment also demonstrate to the recovery based on the local vEPCb
deployed in 5GIC testbed.
 Reliable Slicing in 5G Networks 9

Figure 4: RAN and Core network deployment in federated environment

Figure 5: Proposed scheme experimental evaluation setup

2.2 Service chaining reliability

The advent of SDN and NFV enables a convergent network-cloud ecosystem

offering more effective and operative network and service deployments on top
of virtual networking, computing and storage resources integrated into one
programmable and unified infrastructure while serving a multitude of
distributed smart devices and applications (e.g., robots, drones, smart
vehicles) in turn being part of the infrastructure itself.
 10 Valcarenghi, Giorgetti, Martini, Kondepu, Gharbaoui, Castoldi

Thanks to softwarization, a scenario can be envisioned where service

providers may offer not only communication services, but also virtualized
computing and storage capabilities by elastically slicing the (cloud and
network) infrastructure into partitions (i.e., network slices) offering
customized network functions and services (e.g., NAT, firewall, deep packet
inspection) tailored for specific applications. Moreover, with the
softwarization of telecommunication infrastructures, a new breed of
applications can be conceived (e.g., cloud robotics, smart cities) by
dynamically composing (i.e., chaining) computational and network services
deployed as virtual functions (VFs) in distributed micro-clouds located at the
Edge of the current telecommunication infrastructure. Indeed, SDN can
effectively provide programming abstractions that can be exploited for the
dynamic enforcement and in-line steering of data traffic along the network
path of service chains (i.e., service chain paths).

Through slicing and dynamic service chaining, service providers can deploy
service infrastructures to serve many different verticals while saving capital
and operating expenditures (CAPEX and OPEX). However, the concurrent
usage of resources, the high dynamicity of services and the geographical
distribution of VFs pose new challenges to service providers in terms of
service lifecycle management and automation to address the QoS and service
availability requirements of heterogeneous applications. To this purpose,
close control loops and techniques are required towards providing automation,
resource usage optimization and reliability eventually leveraging network
analytics assisted decisions [27]. In this direction, ONAP is working on new
solutions for providing automation, performance optimization and, in general,
service lifecycle management capabilities [28]. On the other hand, the
reliability of service chains is stated as a primary requirement to assure proper
service availability [29][30]. However, the problem to address QoS and
service chaining reliability is challenging due many different and
heterogeneous application requirements. A way to effectively address
reliability is to assure adaptive resource provisioning and protection
mechanisms while service chains runs aiming at preventing service
degradations due to the concurrent use of resources from different applications
[31][32]. Moreover, application-oriented mechanisms are desirable that can
be achieved through intent-based approach [33] and detection of service
degradation on end-to-end basis [32].
 Reliable Slicing in 5G Networks 11

References
1. https://datatracker.ietf.org/meeting/99/materials/slides-99-netslicing-alex-galis-netslicing-terms-
and-systems
2. NGMN Alliance, “Description of Network Slicing Concept”, NGMN 5G P1 Requirements &
Architecture Work Stream End-to-End Architecture, v1.0.8, Sep. 14, 2016,
https://www.ngmn.org/fileadmin/user_upload/161010_NGMN_Network_Slicing_framework_v1.0.
8.pdf (accessed Apr. 30, 2018)
3. “3rd Generation Partnership Project; Technical Specification Group Services and System Aspects;
Telecommunication management; Study on management and orchestration of network slicing for next
generation network (Release 15)”, 3GPP TR 28.801 V15.1.0 (2018-01)
4. ETSI GR NFV-EVE 012, “Network Functions Virtualisation (NFV) Release 3; Evolution and
Ecosystem; Report on Network Slicing Support with ETSI NFV Architecture Framework” v3.1.1,
December 2017.
5. https://datatracker.ietf.org/wg/netslicing/about/ (accessed april 30, 2018)
6. L. Qiang et al., “Technology Independent Information Model for Network Slicing draft-qiang-coms-
netslicing-information-model-02, https://tools.ietf.org/html/draft-qiang-coms-netslicing-
information-model-02, Expiration date: July 30, 2018 (accessed April 30, 2018)
7. “Applicability of Abstraction and Control of Traffic Engineered Networks (ACTN) to Network
Slicing draft-king-teas-applicability-actn-slicing-02”, https://tools.ietf.org/html/draft-king-teas-
applicability-actn-slicing-02, Expiration date: July (accessed April 30, 2018)
8. https://www.broadband-forum.org/5g
9. ONF, “TR-526 Applying SDN Architecture to 5G Slicing”, April 2016
10. Yoshinori Goto, “Activity Report of ITU-T Focus Group on IMT-2020”, NTT Technical Review,
Vol. 15 No. 6 June 2017
11. http://www.mef.net/Assets/Technical_Specifications/PDF/MEF_55.pdf (accessed April 30, 2018)
12. https://www.mef.net/Assets/White_Papers/MEF_Third_Network_LSO_Vision_FINAL.pdf
(accessed April 30, 2018)
13. https://www.gsma.com/futurenetworks/5g/introduction-to-5g-network-slicing/ (accessed April 30,
2018)
14. 5G-Transformer website, http://5g-transformer.eu/
15. 5G!Pagoda, “D2.3: Initial report on the overall system architecture definition”,
https://goo.gl/2Z7PWu, June 30, 2017
16. SONATA, “D2.2 Architecture Design”,
17. 5G PPP, “View on 5g architecture,” White Paper, https://5g-ppp.eu/white-papers (accessed
10/07/2017).
18. V. G. Nguyen et al., "SDN/NFV-Based Mobile Packet Core Network Architectures: A Survey," in
IEEE Communications Surveys & Tutorials, vol. 19, no. 3, pp. 1567-1602, third quarter 2017.
19. 3GPP TR 38.801, “Study on new radio access technology: Radio access architecture and interfaces
(Release 14)”, V14.0.0 (2017-03).
20. 3GPP TS 23.007, “Technical Specification Group Core Network and Terminals; Restoration
procedures”, Dec 2017.
21. L. Valcarenghi, et al., “Quality-of-service-aware fault tolerance for grid-enabled applications”,
Optical Switching and Networking, Volume 5, Issues 2–3, 2008.
22. ETSI GS NFV-REL 002 V1.1.1 (2015-09), “Network Functions Virtualisation (NFV); Reliability;
Report on Scalable Architectures for Reliability Management”.
23. F. F. Moghaddam, et al., "Self-Healing Redundancy for OpenStack Applications through Fault-
Tolerant Multi-Agent Task Scheduling," IEEE CloudCom 2016
24. EU SoftFIRE project, https://www.softfire.eu/
25. M. Gharbaoui, C. Contoli, G. Davoli, G. Cuffaro, B. Martini, F. Paganelli, W. Cerroni, P. Cappanera,
P. Castoldi, “Experimenting latency-aware and reliable service chaining in Next Generation Internet
testbed facility,” IEEE Conference on Network Function Virtualization and Software Defined
Networks (NFV-SDN), 2018.
 12 Valcarenghi, Giorgetti, Martini, Kondepu, Gharbaoui, Castoldi

26. K. Antevski, et al., “Resource Orchestration of 5G Transport Networks for Vertical Industries,” arXiv
preprint arXiv:1807.10430, 2018.
27. N. Sousa, et al., “Network Service Orchestration: A Survey, ” 2018.
28. http://github.com/onap/clamp
29. A. Hmaity, et al., "Virtual Network Function placement for resilient Service Chain provisioning,"
8th International Workshop on Resilient Networks Design and Modeling (RNDM), Halmstad, 2016.
30. M. Gharbaoui, C. Contoli, G. Davoli, G. Cuffaro, B. Martini, F. Paganelli, W. Cerroni, P. Cappanera,
P. Castoldi, “Demonstration of Latency-Aware and Self-Adaptive Service Chaining in 5G/SDN/NFV
infrastructures,” IEEE Conference on Network Function Virtualization and Software Defined
Networks (NFV-SDN), 2018.
31. AA. Mohammed, M. Gharbaoui, B. Martini, F. Paganelli, P. Castoldi, “SDN controller for network-
aware adaptive orchestration in dynamic service chaining,” IEEE NetSoft Conference and Workshops
(NetSoft), Seoul, 2016.
32. M. Gharbaoui, S. Fichera, P. Castoldi, B. Martini, “Network orchestrator for QoS-enabled service
function chaining in reliable NFV/SDN infrastructure,” ," IEEE NetSoft Conference and Workshops
(NetSoft), 2017.