(Network Slicing on the way)

March 22, 2018

By Thomas Tovinger, 3GPP SA5 Chairman

In December 2017, 3GPP passed two major milestones for 5G by approving the first set of 5G NR specs and by putting in place the 5G Phase 1 System Architecture. These achievements have brought about the need for new management standards, as 5G adds to the ever-growing size and complexity of telecom systems.

3GPP management standards from working group SA5 are approaching another major milestone for 5G. With our studies on the 5G management architecture, network slicing and charging completed last year, we are now well under way with the normative work for the first phase in 3GPP Release 15, which includes building up a new service-oriented management architecture and all the necessary functionalities for management and charging for 5G networks.

SA5’s current work also includes several other work/study items such as management of QoE measurement collection and new technologies for RESTful management protocols. However, this article will focus on the new 5G Rel-15 architecture and the main functionalities, including charging.

5G networks and network slicing

Management and orchestration of 5G networks and network slicing is a feature that includes the following work items: management concept and architecture, provisioning, network resource model, fault supervision, assurance and performance management, trace management and virtualization management aspects. With the output of these work items, SA5 provides specified management interfaces in support of 5G networks and network slicing. An operator can configure and manage the mobile network to support various types of services enabled by 5G, for example eMBB (enhanced Mobile Broadband) and URLLC (Ultra-Reliable and Low Latency Communications), depending on the different customers’ needs. The management concept, architecture and provisioning are being defined in TS 28.530, 28.531, 28.532 and 28.533. 

Network slicing is seen as one of the key features for 5G, allowing vertical industries to take advantage of 5G networks and services. 3GPP SA5 adopts the network slice concept as defined in SA2 and addresses the management aspects. Network slicing is about transforming a PLMN from a single network to a network where logical partitions are created, with appropriate network isolation, resources, optimized topology and specific configuration to serve various service requirements.

As an example, a variety of communication service instances provided by multiple Network Slice Instances (NSIs) are illustrated in the figure below. The different parts of an NSI are grouped as Network Slice Subnets (e.g. RAN, 5GC and Transport) allowing the lifecycle of a Network Slice Subnet Instance (NSSI) to be managed independently from the lifecycle of an NSI.

Provisioning of network slice instances

The management aspects of a network slice instance can be described by the four phases:

1) Preparation: in the preparation phase the network slice instance does not exist. The preparation phase includes network slice template design, network slice capacity planning, on-boarding and evaluation of the network slice requirements, preparing the network environment and other necessary preparations required to be done before the creation of a network slice instance.

2) Commissioning: provisioning in the commissioning phase includes creation of the network slice instance. During network slice instance creation all needed resources are allocated and configured to satisfy the network slice requirements. The creation of a network slice instance can include creation and/or modification of the network slice instance constituents.

3) Operation: includes the activation, supervision, performance reporting (e.g. for KPI monitoring), resource capacity planning, modification, and de-activation of a network slice instance. Provisioning in the operation phase involves activation, modification and de-activation of a network slice instance.

4) Decommissioning: network slice instance provisioning in the decommissioning phase includes decommissioning of non-shared constituents if required and removing the network slice instance specific configuration from the shared constituents. After the decommissioning phase, the network slice instance is terminated and does not exist anymore.

Similarly, provisioning for a network slice subnet instance (NSSI) includes the following operations:

• Create an NSSI;
• Activate an NSSI;
• De-active an NSSI;
• Modify an NSSI;
• Terminate an NSSI.

Roles related to 5G networks and network slicing

The roles related to 5G networks and network slicing management include: Communication Service Customer, Communication Service Provider (CSP), Network Operator (NOP), Network Equipment Provider (NEP), Virtualization Infrastructure Service Provider (VISP), Data Centre Service Provider (DCSP), NFVI (Network Functions Virtualization Infrastructure) Supplier and Hardware Supplier.

Depending on actual scenarios:

• Each role can be played by one or more organizations simultaneously;
• An organization can play one or several roles simultaneously (for example, a company can play CSP and NOP roles simultaneously).

Management models for network slicing

Different management models can be used in the context of network slicing.

1) Network Slice as a Service (NSaaS): NSaaS can be offered by a CSP to its CSC in the form of a communication service. This service allows CSC to use and optionally manage the network slice instance. In turn, this CSC can play the role of CSP and offer their own services (e.g. communication services) on top of the network slice instance. The MNSI (Managed Network Slice Instance) in the figure represents a network slice instance and CS represents a communication service.

2) Network Slices as NOP internals: network slices are not part of the CSP service offering and hence are not visible to CSCs. However, the NOP, to provide support to communication services, may decide to deploy network slices, e.g. for internal network optimization purposes.

Management architecture

SA5 recognizes the need for automation of management by introducing new management functions such as a communication service management function (CSMF), network slice management function (NSMF) and a network slice subnet management function (NSSMF) to provide an appropriate abstraction level for automation.

The 3GPP SA5 management architecture will adopt a service-oriented management architecture which is described as interaction between management service consumer and management service provider. For example, a management service consumer can request operations from management service providers on fault supervision service, performance management service, provisioning service and notification service, etc.

Network Resource Model (NRM) for 5G networks and network slicing

To support management and orchestration of 5G networks, the Network Resource Model (NRM) representing the manageable aspects of 5G networks needs to be defined, according to 5G network specifications from other 3GPP working groups as well as considering requirements from 5G management architecture and operations.

The 5G NRM specifications family includes 4 specifications: TS 28.540 and TS 28.541 for NRM of NR and NG-RAN, TS 28.542 and TS 28.543 for NRM of 5G core network.

According to content categorization, 5G NRM specifications can be divided into 3 parts:

• Requirements, also known as stage 1,
• Information Model definitions also known as stage 2, and
• Solution Set definitions also known as stage 3.

Identified in the specifications of 5G NRM requirements (TS 28.540 and TS 28.542), the NRM of 5G network comprises NRM for the 5G core network (5GC) and NRM for 5G radio access network (i.e. NR and NG-RAN). The 5GC NRM definitions support management of 5GC Network Functions, respective interfaces as well as AMF Set and AMF Region. The NR and NG-RAN NRM definitions cover various 5G radio networks connectivity options (standalone and non-standalone radio node deployment options) and architectural options (NR nodes with or without functional split).

The 5G Information Model definitions specify the semantics and behavior of information object class attributes and relations visible on the 5G management interfaces, in a protocol and technology neutral way (UML as protocol-neutral language is used). The 5G Information Model is defined according to 5GC, NR and NG-RAN specifications. For example, in 3GPP TS 38.401, the NR node (gNB) is defined to support three functional split options (i.e. non-split option, two split option with CU and DU, three split option with CU-CP, CU-UP and DU), so in the NR NRM Information Model, corresponding Information Object Class (IOC) is defined for each network function of gNB specified, and different UML diagrams show the relationship of each gNB split option respectively. Further, in the 5G Information Model definitions, the existing Generic NRM Information Service specification (TS 28.622) is referenced to inherit the attributes of generic information object classes, and the existing EPC NRM Information Service specification (TS 28.708) is referenced for 5GS / EPS interworking relationships description.

Finally, NRM Solution Set definitions map the Information Model definitions to a specific protocol definition used for implementations. According to recommendation from TR 32.866 (Study on RESTful based Solution Set), JSON is expected to be chosen as data modelling language to describe one 5G NRM Solution Set.

Fault Supervision of 5G networks and network slicing

Fault Supervision is one of the fundamental functions for the management of a 5G network and its communication services. For the fault supervision of 5G networks and network slicing, the following 3GPP TSs are being specified:

1) TS 28.545 “Management and orchestration of networks and network slicing; Fault Supervision (FS); Stage 1”, which includes:

• The use cases and requirements for fault supervision of 5G networks and network slicing.
• The definitions of fault supervision related management services (e.g. NetworkSliceAlarmAcknowledgement, NetworkSliceAlarmListReading, NetworkSliceAlarmClearance, NetworkSliceAlarmNotification, NetworkSliceAlarmSubscription, etc.)

2) TS 28.546 “Management and orchestration of networks and network slicing; Fault Supervision (FS); Stage 2 and stage 3”, which includes the definition of:

• Interfaces of the fault supervision related management services; (Stage 2)
• Notifications; (Stage 2)
• Alarm related information models (e.g. alarmInformation, alarmList, etc.); (Stage 2)
• Solution set(s) (e.g. RESTful HTTP-based solution set for Fault Supervison); (Stage 3)
• New event types and probable causes if necessary. 

Assurance data and Performance Management for 5G networks and network slicing

The 5G network is designed to accommodate continuously fast increasing data traffic demand, and in addition, to support new services such as IoT, cloud-based services, industrial control, autonomous driving, mission critical communications, etc. Such services may have their own performance criteria, such as massive connectivity, extreme broadband, ultra-low latency and ultra-high reliability.

The performance data of the 5G networks and NFs (Network Functions) are fundamental for network monitoring, assessment, analysis, optimization and assurance. For the services with ultra-low latency and ultra-high reliability requirements, any faults or performance issues in the networks can cause service failure which may result in serious personal and property losses. Therefore, it is necessary to be able to collect the performance data in real-time (e.g., by performance data streaming), so that the analytic applications (e.g., network optimization, SON, etc.) could use the performance data to detect any network performance problems, predict the potential issues and take appropriate actions quickly or even in advance.

For network slicing, the communication services are provided on top of the end-to-end network slice instances, so the performance needs to be monitored from end-to-end point of view.

The end to end performance data of 5G networks (including sub-networks), NSIs (Network Slice Instances) and NSSIs (Network Slice Subnet Instances) are vital for operators to know whether they can meet the communication service requirement.

The performance data may be used by various kinds of consumers, such as network operator, SON applications, network optimization applications, network analytics applications, performance assurance applications, etc. To facilitate various consumers to get their required performance data, the following items are being pursued by this WI:

• A service based PM framework and a list of PM services as described in the table below:  

Management service name	Management service description
NF measurement job control service	The management service for creating and terminating the measurement job(s) for the NF(s).
NF measurement job information service	The management service for querying the information of the measurement job(s) for the NF(s).
NF performance data file reporting Service	The management service for reporting the NF performance data file.
NF performance data streaming service	The management service for providing streaming of NF performance data.
NSSI measurement job control service	The management service for creating and terminating the measurement job(s) for the NSSI(s).
NSSI measurement job information service	The management service for querying the information of the measurement job(s) for the NSSI(s).
NSSI performance data file reporting Service	The management service for reporting the NSSI performance data file.
NSSI performance data streaming service	The management service for providing streaming of NSSI performance data.
NSI measurement job control service	The management service for creating and terminating the measurement job(s) for the NSI(s).
NSI measurement job information service	The management service for querying the information of the measurement job(s) for the NSI(s).
NSI performance data file reporting Service	The management service for reporting the NSI performance data file.
NSI performance data streaming service	The management service for providing streaming of NSI performance data.
Network measurement job control service	The management service for creating and terminating the measurement job(s) to collect the network performance data that is not specific to network slicing.
Network measurement job information service	The management service for querying the information of the measurement job(s) to collect the network performance data that is not specific to network slicing.
Network performance data file reporting service	The management service for reporting the network performance data file that is not specific to network slicing.
Network performance data streaming service	The management service for providing network performance data streaming that is not specific to network slicing.

• Performance measurements (including the data that can be used for performance assurance) for 3GPP NFs;
• End to end KPIs, performance measurements (including the data that can be used for performance assurance) for NSIs, NSSIs and networks (where the performance data is not specific to network slicing).

Management and virtualization aspects of 5G networks

For 5G networks, it is expected that most of the network functions will run as software components on operators' telco-cloud systems rather than using dedicated hardware components. Besides the virtualization for Core Network (including 5GC, EPC and IMS), the NG-RAN architecture is being defined with functional split between central unit and distributed unit, where the central unit can also be virtualized.

SA5 has conducted a study on management aspects of the NG-RAN that includes virtualized network functions, and has concluded in TR 32.864 that the existing specifications (related to management of mobile networks that include virtualized network functions) need some enhancements for 5G. The enhancements are mainly on the interactions between 3GPP management system and external management systems (e.g., ETSI NFV MANO) for the following aspects:

• Management requirements and architecture;
• Life Cycle Management (e.g., PNF management);
• Configuration Management;
• Performance Management;
• Fault Management.

There are gaps identified between 3GPP SA5 requirements and ETSI ISG NFV solutions in terms of the required enhancements, and 3GPP SA5 is in cooperation with ETSI ISG NFV to solve these gaps.

Although the need for enhancements were found in the study for 5G, SA5 has reached the conclusion that they can be the used for 4G as well. So the specifications for management of mobile networks that include virtualized network functions are being made generally applicable to both 4G and 5G networks.

Study on energy efficiency of 5G networks

Following the conclusions of the study on Energy Efficiency (EE) aspects in 3GPP Standards, TSG SA#75 recommended initiating further follow-up studies on a range of energy efficiency control related issues for 5G networks including the following aspects:

• Definition and calculation of EE KPIs in 3GPP Systems
• Energy Efficiency control in 3GPP Systems
• Coordinated energy saving in RAN and other subsystem in 3GPP Systems
• Power consumption reduction at the site level
• Energy Efficiency in 3GPP systems with NFV
• Energy Efficiency in Self-Organizing Networks (SON). 

TR 32.972 (Study on system and functional aspects of energy efficiency in 5G networks) aims to:

• Identify EE KPI definitions made by ETSI TC EE, ITU-T SG5, ETSI NFV ISG, etc., which are relevant for 5G networks, in addition to definitions made in SA TR 21.866. Such EE KPIs can be defined at various levels, incl. network and equipment levels (potentially, at virtualized network function and virtualized resource level), and per deployment scenario (dense urban, rural, etc.). With 5G, potentially, EE KPIs can be defined at network slice level;
• Identify metrics to be defined by 3GPP so as to be able to calculate the above EE KPIs for 5G networks. Such metrics might relate to data volumes, coverage area or energy consumption;
• Assess whether existing OA&M mechanisms enable to control and monitor the identified metrics. In particular, check if the IRP for the control and monitoring of Power, Energy and Environmental (PEE) parameters for Radio Access Networks (RAN) (TS 28.304, 28.305, 28.306) can be applied to 5G networks. If not, identify potential new OA&M mechanisms;
• Elaborate further on the EE control framework defined in TR 21.866 and identify potential gaps with respect to existing management architectures, incl. SON and NFV based architectures;
• Examine whether new energy saving functionalities might enable the 3GPP management system to manage energy more efficiently. In particular, the applicability of ETSI ES 203 237 (Green Abstraction Layer; Power management capabilities of the future energy telecommunication fixed network nodes) to the management of 5G networks is to be evaluated;
• Identify potential enhancements in existing standards which could lead to achieving improved 3GPP system-wide energy efficiency.

This study requires interactions with other 3GPP working groups and SDOs working on related topics, including ITU-T SG5, ETSI TC EE, ETSI NFV ISG.

5G Charging system architecture and service based interface

Commercial deployment of the Rel-15 5G System will not be possible without capabilities for Operators to be able to monetize the various set of features and services which are specified in TS 23.501, TS 23.502 and TS 23.503. This is defined under the charging framework, which includes e.g. real-time control of subscriber's usage of 5G Network resources for charging purpose, or per-UE data collection (e.g. for CDRs generation) which can also be used for other purposes e.g. analytics.

SA5 has investigated, during a study period in 2017, on how charging architecture should evolve, which key features should be specified as part of charging capabilities, and which alternative amongst charging solutions should be selected, to better support the first commercial 5G system deployment. Based on the study results, the charging architecture evolution and selected Rel-15 key functionalities for 5G system are under ongoing normative phase through development of a complete set of specifications (architecture, functionalities and protocols) A brief overview of the charging coverage for the Rel-15 5G system is provided in this article.

Service Based Interface

One key evolution of the charging architecture is the adoption of a service based interface integrated into the overall 5G system service based architecture, enabling deployments of charging functions in virtualized environment and use of new software techniques. The new charging function (CHF) introduced in the 5G system architecture, as shown in the picture below, allows charging services to be offered to authorized network functions. A converged online and offline charging will also be supported. 

While offering the service based interface to the 5G system, the overall converged charging system will be able to interface the billing system as for the existing system (e.g. 4G) to allow Operators to preserve their billing environment. These evolutions are incorporated in the TS 32.240 umbrella architecture and principles charging specification. The services, operations and procedures of charging using Service Based Interface will be specified in a new TS 32.290, and TS 32.291 will be the stage 3 for this interface.

5G Data connectivity charging

The "5G Data connectivity charging", achieved by SMF invocation of charging service(s) exposed by the charging function (CHF), will be specified in a new TS 32.255, encompassing the various configurations and functionalities supported via the SMF, which are highlighted below.

For 3GPP network deployments using network slicing, by indicating to the charging system which network slice instance is serving the UE during the data connectivity, the Operator will be able to apply business case charging differentiation. Further improvements on flexibility in charging systems deployments for 5G network slicing will be explored in future releases.

The new 5G QoS model introduced to support requirements from various applications in data connectivity, is considered to support QoS based charging for subscriber's usage. 5G QoS-based charging is also defined to address inter-Operator's settlements (i.e. between VPLMN and HPLMN) in roaming Home-routed scenario.

All charging aspects for data services in Local breakout roaming scenarios will be further considered.

In continuation with existing principles on Access type traffic charging differentiation, the two Access Networks (i.e. NG-RAN and untrusted WLAN access) supported in Rel-15 are covered.

Charging capabilities encompass the various functionalities introduced in the 5G system to support flexible deployment of application functions (e.g. edge computing), such as the three different Session and Service Continuity (SSC) modes and the Uplink Classifiers and Branching Points.

Charging continuity for interworking and handover between 5G and existing EPC is addressed.

In 5G Multi-Operator Core Network sharing architecture (i.e. shared RAN), identification of the PLMN that the 5G-RAN resources were used for to convey the traffic, allows settlements between Operators.

The stage 3 for "5G data connectivity charging" will be available in TS 32.298 for the CDRs’ ASN.1 definition and in TS 32.291 for the data type definition in the protocol used for the service based interface.

Acknowledgements

The author wishes to thank the SA5 Rapporteurs of the indicated work items for their input to this article: Zou Lan, Jan Groenendijk, Zhang Kai, Weixing Wang, Zhu Weihong, Yizhi Yao, Xiaowen Sun, Jean-Michel Cornily, Maryse Gardella and Chen Shan. Thanks also to Kevin Flynn for help with graphics.