Network Slicing is an advanced mechanism to provide Quality of Service (QoS) in 5G networks. With Network Slicing, the Mobile Operator will be able to deliver heterogeneous demands and support different requirements by the end customer. The main objective of Network Slicing is to enable the Operator to partition the network resources to allow traffic handling of different users with diverse network requirements. For example, sharing of a given physical network to simultaneously run IoT, MBB, URLLC, V2X applications, with different transmission characteristics. All these characteristics are to be served by a common 5G network platform that has to cater individually for the MBB slice, the IoT slice, and the low-latency slice.
Network Slicing Principles
Network Slicing uses a specific form of virtualization that allows multiple logical networks to run on top of a shared physical network infrastructure. It provides an end-to-end virtual network, including not just networking but compute and storage functions too. Network Slicing is also able to partition the physical system at an end-to-end level to allow optimum grouping of traffic, isolation from other tenants, and configuring of resources at a macro level. In LTE QoS, DiffServ, QCI, VPN, and IP Sec have overlap with the functionality of network slicing. The new approach, however, allows a virtual partition of the network and provides new service and business opportunities in the 5G era, such as Power grid communications, Factory network, Concert streaming with a temporary lease, etc.
Network Slicing in 3GPP
In 3GPP Release 13, the concept of a dedicated Core Network (DECOR) was introduced. With DECOR a device can be directed to a specific EPC of the network. DECOR applies to virtualized environments since no additional Public Land Mobile Network (PLMN) Ids need to be allocated. Also, the Selection and Redirection of the Mobility Management Entity (MME) is based on the subscribed User Equipment usage type. The MME further selects the serving gateway and the Packet Data Network Gateway (SGW/PGW) based on the UE usage type. One UE corresponds to one UE usage type, and there is no indicator in Radio Resource Control (RRC) to let the eNB select the right MME, and all changes are core network-centric.
3GPP Release 14 describes the Enhanced DECOR (eDECOR) concept that introduces UE assisted Dedicated Core Network selection. The UE is provisioned with a default Dedicated Core Network ID (DCN ID) by the Home PLMN. The serving network provides a DCN ID for that PLMN. The UE stores per PLMN DCN ID and the DCN ID are carried in the RRC. The eNB selects the appropriate MME based on the DCN ID. Again, the MME selects SGW/PGW based on the UE usage type, since one UE corresponds to one DCN ID. All PDN connections of the UE are in the same DCN.
In Release 15, a brand new 5G core network and system architecture are introduced. One UE can connect to multiple slices, while the Access and Mobility Management Function (AMF) with is the equivalent of the MME in 5G core, is common to all slices. Each slice is identified by the Specific Network Slice Selection Assistance Information (S-NSSAI). The S-NSSAI contains the Slice Differentiator (SD) and the Slice Service Type (SST) which defines the three basic service types according to IMT2020, eMMB, mMTC, and URLLC. The Network Slice Selection Assistance Information is a set of S-NSSAIs , with maximum 8xS-NSSAIs sent in signaling messages between the UE and the network. The UE is provisioned with a configured NSSAI per PLMN and provided with allowed NSSAI by serving PLMN. The UE can be connected up to 8xS-NSSAls (slices) simultaneously, which means it can be served by up to eight network slices at a time. The requested NSSAI is carried in the RRC during the initial access when 5G Globally Unique Temporary Identity (5G-GUTI) is not available, and the selection of Network Functions (NFs) is based on the S-NSSAI.
Creating a network slice for a customer
As mentioned previously, the need for creating a network slice derives from the customer requirements. The basic principle is that a customer orders a service from the Mobile Operator with specific network requirements. These requirements can be based on the type of service, capacity, latency, performance, coverage, security, etc. The Operator processes the demands and generates the appropriate network slice according to the customer needs.
To receive the appropriate service, the customer provides the Generic Service Template (GST), which holds information on the service type (e.g., URLLC, eMBB, etc.), the performance (e.g. latency requirements, reliability), the capacity and the functionality. The NEtwork Slice Type (NEST) includes specific values of the GST parameters, for example, 99.995% availability, 20msec latency, 10Gbps data rate, etc. The customer provides these service requirements to the Operator, known in this process as the Network Slice Provider (NSP), who, in turn, translates this template to the appropriate Network Slice Template (NST). This way, the network instantiation process is triggered, and the 5G system provides end-to-end network slice management for all network segments, such as 5G RAN, Transport, and Core. At the registration phase, the 5G gNodeB may select the AMF for each requested slice based on the Requested NSSAI or the 5G-GUTI received over the RRC, otherwise, it follows the default AMF. The Unified Data Management (UDM) includes the slice subscription information. At the same time, the Network Slice Selection Function (NSSF) determines the slices at the Access and Core, together with the NF Repository Function (NRF) which is used for service discovery with slice related identifiers.
Once the network slice is created, it is valid within a Registration Area (RA), which consists of one or more Tracking Areas (TAs). While Registration Areas are defined and used in the core network, only Tracking Areas are visible to the 5G RAN. Each Tracking Area (TA) is identified by the TA Identifier (TAI), which is broadcasted in each cell, and not in the network slices. Supported network slices per cell are indirect via TA to RA mapping.
Network Slicing benefits
Using the above procedures developed by 3GPP, Network Slicing is becoming the most efficient way to tackle complex service demands and provide superior Quality of Experience to the end-user in a 5G network. Being able to adapt its resources according to specific customer needs, it provides flexibility and scalability to support the wide variety of 5G services using a common network platform. This common platform, based on an end-to-end 5G system in RAN, transport, and core will be able to help the Mobile Operators deal with diverse 5G service requirements and cope with the growing demand for over-the-top (OTT) applications and services.
