5G networking: non-public networks & network slicing
A significant aspect of fifth-generation (5G) networking is its support for the internet of things (IOT). The IOT broadly refers to the increasing integration of different classes of devices within wireless networks. As part of this drive towards universal connectivity, an ever-growing range of smart devices is now available - from watches, to cars, to household appliances – all configured to seamlessly connect to online applications via wireless telecommunications.
In parallel with the growth of the IOT, the advent of 5G networking is drastically increasing data transfer rates and reducing congestion and latency, leading to a dramatic increase in the density of devices which can be supported in a given network.
5G capabilities are rapidly expanding the range of IOT (and IIOT) use cases which can be viably supported.
Industrial internet of things
An important subset of the IOT is the industrial internet of things (IIOT), in which networked sensors, robots, user terminals, actuators and other devices are provided with wireless connectivity to support new use cases within industry. A key distinguisher of the IIOT over the broader IOT is the consequence of failure. Whereas much of the IOT is concerned with non-critical applications such as fitness trackers and household appliances, where congestion, latency, reliability and system failure are not of significant consequence, such factors may be mission-critical in the IIOT. In domains such as oil and gas, manufacturing, power generation, and healthcare, communication delays or system failures may at best have significant economic implications, and at worst be endangering to life.
Exemplary IIOT use cases include the use of networked robots for performing surgery or manufacturing tasks, the use of networked sensing arrays to monitor industrial processes (for example the operation of an oil refinery or nuclear power station), and the use of networked microphones, cameras and monitors for live television broadcasting. In common with many use cases in the IIOT, each of these use cases involves stringent service requirements, such as low latency, and / or high data throughput, reliability, security and quality of service (QoS).
Ultra-reliable low-latency communication
In order to meet such requirements, the 5G standard supports features including ultra-reliable low-latency communication (URLLC) for time-critical use cases such as robotic surgery and vehicle-to-vehicle communication; enhanced mobile broadband (eMBB) for applications requiring high bandwidth internet access (for example, virtual reality), and massive machine type communication (mMTC) to extend internet access to high-density arrays of low-power sensors, meters, and remote monitoring devices.
Public land mobile networks (PLMNs)
In general, 5G telecommunication systems will be deployed as public land mobile networks (PLMN), under the operation of a mobile network operator, providing simultaneous support for a large ecosystem of devices spanning a wide variety of use cases. However, for many IIOT use cases, particularly those with significant security, reliability and flexibility requirements, the use of a PLMN may be unattractive. For example, in a use case such as robotic-assisted surgery, varying QoS resulting from wider traffic flows on a PLMN may be highly detrimental to reliability and latency, both of which are critical safety requirements. In use cases such as a nuclear power station or military installation, where security is a significant concern, the transmission of sensitive data over a PLMN may be considered inappropriate. Historically, such use cases have gravitated towards the use of distinct local networks, for example a private network implemented using Ethernet, Wi-Fi, or Bluetooth protocols.
Non-public networks (NPNs)
With the advent of 5G comes support for non-public networks (NPN), which seek to combine the security and configurability benefits of a private network with the significant advantages of 5G (for example, support for URLLC, eMBB and mMTC).
The use of an NPN brings a number of distinct benefits, including:
- Further reliability and latency gains (by siting 5G network functions and service applications as close as possible to the devices supported by the network).
- Full configurability (allowing the operator to optimise all aspects of network operation to a specific usage case)
- Integrity and accountability of traffic flows (providing enhanced security).
- Isolation from the public domain (reducing the risk of malicious attacks and exposure to PLMN system outages).
- Independent operation (allowing the operator complete control over authorisation and authentication of devices in the network).
Stand-alone non-public networks (SNPNs)
A key feature of how 5G implements NPNs, when compared to classical private networks, is the flexibility in the degree of integration between the NPN and one or more PLMNs providing coverage to the NPN site.
At one end of the spectrum is a truly stand-alone NPN (SNPN), defined by its physical separation from any of the network functions usually provided by a PLMN. Accordingly a SNPN uses physically distinct radio resources, subscriber database, and dedicated hardware to support both the radio-access network (RAN) and the core network (CN) components of the system. In this sense, an SNPN is in effect a scaled-down PLMN with a restricted subscriber set. A SNPN may link to a PLMN via an edge node with a firewall (to provide access to, for example, voice services), but in the most stringently separated deployments, user equipments (UEs) within the SNPN will not communicate directly with the PLMN over the air interface.
PLMNs, NPNs and SNPNs
For some use cases in the IIOT, the benefits of independence from a PLMN may justify the overheads of setting up and maintaining a standalone SNPN.
In many instances, a degree of integration between an NPN and one or more PLMNs may be advantageous.
In a first case, RAN hardware supporting the NPN may be shared with a PLMN, while other network functions of the NPN remain physically separate from the PLMN (that is, the NPN provides its own user plane gateway, core network and subscriber database).
In a second case, the NPN again shares the RAN with a PLMN, but further relies on the PLMN to provide control plane functions and the subscriber database.
In a third case, the NPN is fully hosted by a PLMN, in that the PLMN provides all the network functions of the NPN. In this latter scenario, public network traffic and NPN network traffic are treated as distinct parts of the PLMN, an approach achieved via virtualisation of network functions, as supported by the 5G standard.
A benefit of this latter approach is that NPN subscribers are by definition subscribers of the PLMN which hosts the NPN, enabling roaming of NPN UEs (for example, user terminals or vehicles) in the wider PLMN to be easily implemented.
Where the NPN is fully or partially integrated with a PLMN, 5G enables logical separation of the NPN and public network functions of the PLMN, in order to enable delivery of a required QoS for the NPN, which, as set out above, may be essential for meeting the often stringent latency and reliability requirements imposed by IIOT use cases.
Where the NPN is integrated with a PLMN, dedicated network slicing, also supported by 5G, can provide an additional element of security not afforded by conventional network protocols used to support private networks (for example, WiFi), by maintaining complete logical distinction between NPN and public traffic within the PLMN.
The optimal degree of separation between the NPN and the PLMN, ranging from full-physical separation at one extreme (a SNPN) and full integration with a PLMN at the other, with partial integration at varying degrees of physical and / or logical separation in the middle, is a trade-off between a variety of factors. Where significant flexibility is required to create, configure, scale, operate and monitor network functions, such as in an automated manufacturing context (such as a factory), this may only be achievable via a physically distinct SNPN. This is because individual use cases in the IIOT may have highly divergent requirements to those for which a standard PLMN is optimised. Conversely, the significant overheads involved in establishing and maintaining an SNPN may mitigate towards an NPN fully or partially integrated with a PLMN.