Network Slicing

Source: https://www.linkedin.com/pulse/netwo...utosh-kaushik/

Introduction:

Well I leave up to you to think & explore your imaginations for a convincing answer to the title question and proceed on to technically introduce what a Network Slicing is.

Network Slicing is about transforming the network/system from a static "one size fits all" paradigm, to a new paradigm where logical networks partitions are created, with appropriate isolation, resources and optimized topology to serve a particular purpose or service category or even individual customers. I liked this simple definition from GSMA –

“Network slicing is the embodiment of the concept of running multiple logical networks as virtually independent business operations on a common physical infrastructure in an efficient and economical way.”

Expected to have more potential for Business Customers’ use cases, Network slicing offers an effective way to meet the requirements of a multitude of services and applications over a common network infrastructure, including smartphones, tablets, AR/VR live broadcast, personal health devices, critical remote-controlled equipment, automotive connectivity, industry automation as few examples. Hence the role of network slicing is to support very diverse and extreme requirements for latency, throughput, capacity and availability.

Network slicing will create end-to-end logical networks that have isolated properties and are operated independently. As new services get layered onto the network, a cloud-native core for example will be able to create an instance, or slice, of an entire network virtually. The slice will be fully customized with network resources (dedicated if needed) allocated by use case, subscriber type or application from a common infrastructure.

The operational aspects of network slicing can be considered as three different layers; Service instance layer (e.g. eMBB, URLLC and massive IoT), Network slice instance layer (provides network characteristics required by service instance) and Resources layer. The resources layer has three different subcategories; Access network resources (RAN or fixed network), transport/SDN resources and core network or cloud/VNF resources.

Each network slice is made of subnets viz. RAN subnet, CN subnet, Transport subnet. NFV, SDN, cloud technologies & automation are its enablers.

Network slicing can create revenue opportunities for operators in few ways:

• Serving existing high-end customers with higher grade service: services tailored to the needs of a current customer or customer segment are more valuable and can win a premium Average Revenue Per User (ARPU)

• Reaching new customers for whom premium service is critical: new customer segments can be addressed, creating completely new, additional revenue streams

Benefits of Network slicing from a Network Operator perspective can be:

• Deliver a mix of services with very diverse needs in bandwidth, latency, number of devices, coverage, reliability, … with a common physical network

• Enable custom behaviors and separation/isolation between multiple tenants

• Efficiently utilize mobile network resources – compute, transport, radio – to meet needs of diverse services and tenants

• SLAs based on attributes beyond the usual KPIs of BW / bytes consumed. Expose to customers flexible levels of visibility, control, management of network functions.

• Target customer segments/groups – enterprise, verticals, public safety. Enable customized service offerings to each.

Network Slicing Deployments happening today (with EPC):

∘ Dedicated Core for M2M.

∘ Network Sharing / MVNO.

∘ EPC overlay for 5G, FWA.

Network Slicing Deployments in the future (with 5GC):

∘ To support new business models –

• Dynamic Network Slice as a Service (NSaaS) with guaranteed SLAs to enterprise and vertical tenants

• Nation-wide / global slices with guaranteed SLAs (e.g. V2X, Public Safety)

• Rich set of network slice characteristics

Important Terminologies:

Network Slice Instance (NSI) – A set of Network Function instances and the required resources (e.g. compute, storage and networking resources) which form a “deployed” Network Slice.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.

Network Slice Selection Assistance Information (NSSAI) - The NSSAI is a collection of up to 8 S-NSSAIs.

Single Network Slice Selection Assistance information (S-NSSAI) - It identifies a Network Slice. S-NSSAI = Slice/Service type (SST) + Slice Differentiator (SD). The S-NSSAI may be associated with a PLMN (e.g., PLMN ID) and have network-specific values or have standard values. An S-NSSAI is used by the UE in access network in the PLMN that the SNSSAI is associated with.

Slice/Service type (SST) - Refers to the expected Network Slice behaviour in terms of features and services. Standardized SST values provide a way for establishing global interoperability for slicing so that PLMNs can support the roaming use case more efficiently for the most commonly used Slice/Service Types.

Slice Differentiator (SD) - An optional information that complements the Slice/Service type(s) to allow further differentiation for selecting a Network Slice from the potentially multiple Network Slices that all comply with the indicated Slice/Service type.

Network Slice Subnet (NSS) - A representation of the management aspects of a set of Managed Functions and the required resources (e.g. compute, storage and networking resources).

Network slice subnet instance (NSSI) - An instance of Network Slice Subnet.

Network Slice Subnet Template (NSST) - Description of the structure (and contained components) and configuration of the network slice subnet.

Network Slice Selection Function (NSSF) – A VNF that selects set of network slice instances serving the user equipment, determining the allowed NSSAI and possibly mapping subscribed S-NSSAIs and determining AMF set to be used to serve the UE or, based on configuration, a list of AMFs. NSSF is provisioned to keep mapping between an NSSAI and AMF FQDN (Fully Qualified Domain Name) or IP address. NSSF also may use Network Repository Function (NRF) identifier to help resolving an AMF or other functions for a given slice instance.

Communication Service Management Function (CSMF) - The CSMF involves provisioning and management of communication service instances. Part of its role is to request the necessary resources to realize the communication service instances. The request for the resources includes service specific instance information to be used by the resource management to realize the communication service instance. The CSMF is split into two parts where one of them is related to customer aspects and the other is related to the service and resource aspects. CSMF carries out Service assurance and SLA enforcement for each service instance that is in active operation.

Network Slice Management Function (NSMF) - NSMF performs Network slice instance (NSI) monitoring, reporting and LCM i.e. Slice level health monitoring and SLA assurance & Slice life cycle management and closed-loop operations.

Network Slice Subnet Management Function (NSSMF) – NSSMF performs Network slice subnet instance (NSSI) monitoring, reporting and LCM i.e. Alarm correlation and statistics aggregation on slice subnet level & Network slice subnet instance (NSSI) life cycle management and provisioning according to slice profile.

Network Function Management Function (NFMF) – It does Network function (NF) monitoring, reporting and configuration i.e. Raw slicing alarm/counter collection on NF level & NF slicing parameter configuration and provisioning.

Network Slicing Model (3GPP SA5 Perspective):

No alt text provided for this image
Slice Life Cycle management (Example):

No alt text provided for this image
Description:

Design and create the E2E network slice with specified SLAs. Divide E2E slice SLAs into slice subnet SLAs and network resource requirements.
Proceed with network slice subnet instantiation/activation and SLA application through NSSMF and VNFO/VNFM –
2.1 Virtual network resource allocation and instantiation via VNFO&VNFM

2.2 Translating SLAs into NF configuration parameters and provisioning via NFMF

3. Create connectivity between NFs/Data Centers through Transport Subnets (with Transport NSSMF/SDN Controller).

4. Instantiate needed VNFs with basic configurations through VNFM.

5. VNFM triggers VNF integration with NFMF(EM) and new VNFs get managed by NFMF(EM) for FCAPS.

6. NFMF proceed with parameter provisioning to NFs to activate the service and apply SLAs. In this case the NSSAI for this slice is also provisioned to HSS/UDM for those UE that are allowed to access this service.

Network Slicing Operation:

Unlike 4G/LTE slicing, 5G solution offers two slicing types: -

Hard slicing: network slices must be completely isolated from each other.

Soft slicing: network slices may share certain network resources.

Complete slice isolation requires that the given VNF, e.g. AMF can be used only by a specific slice identified by S-NSSAI, or a set of specific slices identified by NSSAI. Slice sharing would permit the given VNF, e.g. AMF to be used by any network slice.

General Design Principles:

AMF - Must be common to all Network Slice Instances serving the UE. Separate AMFs possible for different groups of UEs (e.g. per tenant). Shared usually.

SMF - Typically dedicated to a slice. Dedicated usually.

UPF - Typically dedicated to a slice. Instantiated on edge cloud close to UE for low latency services. Dedicated usually.

PCF - Shared or dedicated to enable slice specific policy and charging control capabilities. Dedicated usually.

NEF - Typically a dedicated NEF would be used to isolate external access to the network slice. Dedicated usually.

NRF - PLMN level, shared between multiples slices or dedicated to a single slice. Assigned by NSSF during registration.

UDM - Global function within PLMN. May be specialized via NRF based discovery.

AUSF - Global function within PLMN. May be specialized via NRF based discovery.

AF - Application functions are typically dedicated to specific functionality provided by the slice. Dedicated usually.

UDR - Global function within PLMN. May be specialized via NRF based discovery.

UDSF - Global function within PLMN. May be specialized via NRF based discovery.

NSSF - Global function within PLMN. Configured to AMFs.

RAN – Can be shared or dedicated depending on implementation & use case . Like RAN CU-UP can be dedicated function located on edge cloud for low latency services.

UE Assisted Network Slice Selection:

Based on the Requested NSSAI (if any) and the Subscription Information, the 5GC is responsible for selection of a Network Slice instance(s) to serve a UE including the 5GC Control Plane and User Plane Network Functions corresponding to this Network Slice instance(s).

The Network Slice instance selection for a UE is normally triggered as part of the registration procedure by the first AMF that receives the registration request from the UE. The AMF retrieves the slices that are allowed by the user subscription and interacts with the Network Slice Selection Function (NSSF) to select the appropriate Network Slice instance (e.g., based on Allowed S-NSSAIs, PLMN ID, etc.). This could result in a change of AMF if needed.

At registration phase, the gNB selects the AMF which supports UE requested slices based on Requested NSSAI (Assistance Information) or 5G-GUTI received over RRC in Connection Setup Complete. Otherwise gNB selects default configurable AMF.

The establishment of User Plane connectivity to a Data Network via a Network Slice instance(s) comprises two steps:

Performing a RM procedure to select an AMF that supports the required Network Slices.
Establishing one or more PDU Session to the required Data network via the Network Slice Instance(s).
UE Registration:

Functional steps:

Configured NSSAI provisioned to UE based on NS Selection Policy (NSSP): Rules associating an App with a S-NSSAI and DNN linked to the App
UE sends Registration Request including Requested NSSAI. RAN selects AMF based on Requested NSSAI + other criteria. Security is performed. In roaming case the Allowed NSSAI includes a mapping of each S-NSSAI to the S-NSSAIs values the UE subscribes to in the HPLMN.
AMF fetches subscription data from UDM. UDM returns (Sub data+ Subscribed NSSAI)
AMF interrogates NSSF for Slice Selection: NSSF returns set of NSI ID, Allowed S-NSSAI
AMF sends Registration Accept/Complete (Allowed S-NSSAIs, 5G-GUTI, Registration area. Mobility restrictions, …) to both RAN and UE (NAS)
Subsequently UE uses updated NSSAI providing list of Allowed S-NSSAIs.
PDU Session Establishment:

A Protocol Data Unit (PDU) session is a 5G concept for an association between the device and a data network, which can be IP, Ethernet or Unstructured (i.e. transparent to the 5G system). The data transmission can take place after a PDU session to a Data Network is established in a Network Slice. The S-NSSAI associated with a PDU Session is provided to the Access Network, and to the policy and charging entities, to apply slice specific policies.

Functional Steps:

UE start APP A which is mapped to S-NSSAI-A. UE sends PDU session establishment for a given DNN and S-NSSAI.
RAN sends PDU session establishment to AMF selected during Registration
AMF gets from NRF the list of SMF ID (IP@/FQDN) for given NSI and selects SMF
AMF sends Nsmf PDU Session Create Req to SMF
SMF checks subs data, polls PCF (not shown) for policy, select UPF based on DNN/load information and N4 Session establishment
SMF responds session accept with S-NSSAI to AMF and RAN
RAN sends a session accepted to UE
Last Bonus Tip : A Million Dollar Question (��)

Network Sharing vs. Slicing vs. APNs vs. QoS Flows:

Network Sharing allows multiple operators to share the same RAN network (“MOCN/PLMN ID”).
Networks slicing allows operator to partitioning his network into multiple network slices via NSSAI (5G) / DCN ID (4G).
Within a network slice, UEs can create PDU sessions to different Gateways via DNNs(5G) / APNs(4G).
Within a PDU session the UE can establish multiple QoS flows(5G) / EPS Bearers(4G) with specific service characteristics.
In summary, Sharing / Slicing / APNs / QoS mechanisms are complementary tools to partition network resources.