Given the hype, you would be forgiven for thinking that the concept of network slicing is completely new. Truth be told, network slicing ‒ at least in the radio path ‒ has been used in 4G for years and was arguably present in earlier generations of mobile as well.
The difference with 5G network slicing is the innovation that happens within the backbone infrastructure and transport layer, which lets operators provide portions of their networks for specific customer use cases – be it a connected car, a smart home or an Internet of Things (IoT) factory.
Carving up a real slice
5G network slicing is every bit the game-changing technology that the prevailing hype suggests. In a true 5G network, the slice extends beyond the radio path into the core network and all the way through to the customer. It is an end-to-end slice that will allow an operator to provide a number of different virtual network services over the same physical infrastructure.
End-to-end 5G network slicing is not just revolutionary in terms of services for consumers and business customers – it’s revolutionary in terms of the way an operator has to manage and control the network. Current network management systems effectively have to deal with only one type of network. In a standalone 5G network, the management systems will need to deal dynamically with many different virtual networks, offering different services, at different speeds, with different guarantees of delivery and different tariffs – atop a shared physical infrastructure.
A true 5G network holds out the promise of IoT networks on a massive scale, where operators will have to manage the demands of billions of devices all looking for a tiny, low-cost but dedicated slice of the network to transmit small amounts of data. At the same time, it will need to offer massive bandwidth to other applications and users such as broadcasters, while simultaneously providing some businesses and consumers with guaranteed ultra-low latency (ULL) connections for interactive services with round-trip response times measured in a few milliseconds.
Slice it as you like
This is where 5G changes the game and makes an impact across a wealth of industries and applications. For example, consumer applications like immersive gaming or 4K video streaming; health market use cases such as remote monitoring of patients or telesurgery; and advances in the automotive industry such as controlling robots on vehicle assembly lines to autonomous cars and driver support systems. The potential impact of 5G on smart city systems is also widespread covering drone surveillance, smart metering and public safety services. All of these applications require reliable wireless connectivity, but have vastly different connection characteristics and make different demands on the network. Dynamic slicing of a 5G network enables operators to create an end-to-end slice tailored to the needs of each application.
These services represent big business for operators looking to recoup their massive investment in 5G. Indeed, the GSMA estimates that network slicing could enable operators to address a revenue opportunity worth some $300 billion by 2025.
Game on with slicing
Unlocking this opportunity requires what some call “smart slicing”. It can intelligently manage the end-to-end lifecycle of each 5G network slice by using advanced AI and machine learning techniques to immediately understand what the user intends to do and allocates the connection to the right type of slice automatically. Smart slicing determines the characteristics required to meet the user’s needs, and all the other factors of a service-level agreement (SLA), such as connection speed, round-trip time and tariffing. This enables the operator to automatically manage the lifecycle of that slice and the individual user’s connection.
Take, for example, cloud gaming: Avid gamers, paying for a network bundle that includes cloud gaming with an ultra-low latency connection, need to be connected ultra-fast via a dedicated slice as soon as that mobile device connects to a service like Google Stadia, the cloud-based games streaming platform. When game time is over, and the consumers are simply checking email and reading news sites, their use of the the slice can be terminated and their connections revert to 5G’s standard eMBB (enhanced Mobile Broadband) service.
These types of handovers between slices require advanced automation techniques. The network management functions can happen millions of times every hour, across an entire network of millions of people and billions of devices.
To manage this effectively, the smart slicing architecture must use a network data analytics function that has been defined by 3GPP for 5G networks. It is an open, non-proprietary standard called the network data analytics function (NWDAF). Together with two other standard 5G functions ‒ covering policy control (PCF) and network slice selection (NSSF) – NWDAF can provide a data-driven approach for network slice lifecycle automation.
Anticipating every move
But to enhance the capabilities of NWDAF even more requires the use of AI and ML. This can provide advanced ML applications to anticipate issues at the network slice level ‒ such as traffic congestion, latency degradation, link failures or intrusion attacks. NWDAF can use this intelligence to enable the PCF and NSSF functions to manage the network efficiently and deliver the best quality of service.
The AI works by analysing and understanding the data gathered to create and share reusable ML models and network automation ‘playbooks’. The models become the brains of the networks to understand what is happening; the playbooks provide trusted, proven workflows that allows the network to automatically manage services to meet customer SLAs.
It is time to get smart
With smart slicing, operators will be able to provide and guarantee true network-as-a-service applications. Smart slicing will enhance operational efficiency and boost revenues by enabling differentiated operator-led services.
Given the potential, now is the time for operators to take a more strategic approach to their customer relationships. Only by getting close to a targeted set of industry verticals can they fully understand the potential use cases to develop and test products and solutions. Put simply, those operators who intimately understand the customer’s requirements will be the ones best placed to provide the smart network slices that can transform their 5G business.
Subhankar Pal is an Assistant Vice President of Technology and Innovation at Altran, the global leader in engineering and R&D services. Prior to Altran, he worked as a Senior Solutions Architect at Nokia.
