5G service assurance

Context is king: rising to the challenge of contextual service assurance

In this piece written for TMN, Yuval Stein, AVP Product Management, Service Assurance at TEOCO says that operators will need to master the art of contextual service assurance (a term that TEOCO coins in its own markering material) in order to satisfy the demands of business and deliver the full promise of 5G.

There’s little doubt that 5G, as a key enabler to support massive scale Internet of Things uptake, presents the service provider community with a golden opportunity and massive challenge; both technically and from a service delivery standpoint. The range of 5G services now being developed stretches from smart tech to control small gadgets in the home, to systems that can guide heavy-load carrying vehicles around vast factory floors; from immersive virtual reality life experiences to wearable media tech that can monitor vital signs and actually save lives.

But realising the benefits of these opportunities will require an approach that fundamentally changes how networks are designed, delivered, managed and accessed. One key challenge will be in what we currently term service assurance.

In telecoms, the term ‘service assurance’ simply used to mean that a connection was available. The approach was very focused on resource management. Today, availability alone is not enough. Service providers also need to guarantee other factors – the speed of the line, its capacity, and the back-up routes and systems that maintain agreed performance standards in the event of the failure of one network component or card. Shortly, however, even that more up-to-date definition of service assurance is about to get more complex.

How, for example, do you manage the network to assure the right level of service is maintained across one type of functionality within a network slice, without adversely affecting performance in another slice at the same time?

A slice of 5G 

One of the promises of 5G is that, via network slicing, it will be capable of offering different levels and types of service over the same infrastructure. These slices will support different customers and devices with different applications and differential tariffing. Each slice and each solution will have very specific requirements in terms of service assurance.

But how, for example, do you manage the network to assure the right level of service is maintained across one type of functionality within a network slice, without adversely affecting performance in another slice at the same time? If the resource is effectively finite, allocating more of it to one service could potentially affect the level of customer satisfaction elsewhere. A 5G network may live and breathe like no other, but there are some laws of physics that remain constant.

Addressing and solving this challenge is a priority for the operators because enterprise customers will be reliant on that reliable, guaranteed, data connectivity across 5G networks to drive their businesses forward. That is why operators need to master the art of contextual service assurance.

Understanding the impact
Contextual service assurance in a 5G network refers to the ability to monitor a horizontal infrastructure running multiple service types that can be dynamically requested, scaled and terminated. This network will also handle such extremely large data volumes – both in terms of customer traffic and network signaling. Traditional service assurance tools and the manual oversight by engineers of the network in the operations center will be neither practical or possible.

Solving the contextual service assurance challenge is fundamental to getting a return on 5G investment

In a cloud-supported 5G network, the impact of one service type must be understood, managed, and adjusted according to its context – for example the importance of the service, its likely duration and performance requirements  – together with the availability of resources and the prevailing network conditions. These parameters must be understood for the applications and services being run by each slice within a single coverage area.

In fact, because the underlying economic case for deploying a 5G network is predicated on its ability to extend and support the co-existence of various service types on an end-to-end basis across multiple network domains, solving the contextual service assurance challenge is fundamental to getting a return on 5G investment.

Managing the slice

Service assurance changes substantially when using virtualised functions, and a network slice is effectively a self-contained virtual network – meaning it is free of many of the constraints of a hardware-dependent network. Each slice must meet both SLA and operating level agreement (OLA) thresholds determined by the provider and the customer. Thereafter, the service management and assurance also needs to be applied on a per slice basis. This is particularly important for a slice that may be event related – resource heavy at times but also frequently dormant – and where context really is king in terms of network management.

So service assurance tools need to be able to correlate service experience with underlying resources and ensure that no slice is consuming more than it needs and that sufficient capacity is available to boost services as and when required. Different services will have different thresholds. Some of these will be hard – in low-latency or ultra-reliable services for example – while others will have more ability to flex with demand. Understanding and managing this co-existence and interplay between the slices lies at the heart of contextual service assurance.

Harnessing AI

That’s where both Machine learning and AI can come to the rescue of the smart operator, extracting value from enormous streams of data collected from multiple diverse sources to supply both predictive and reactive tools to lift service assurance to a higher level. When combined and made easily accessible this reduces the volume of data that needs to be acted upon, and the time and effort it takes detect and fix issues. The value of those analytical tools grows significantly as they become essential elements in process automation, improving operational efficiency and enhancing customer experience.

Machine learning will become essential to identify and analyse patterns of behavior to drive AI assisted network automation and further reduce the need for human oversight for 5G network. Analytics modules such as Machine-Learning Root Cause Analysis (MC-RCA), will introduce a leap forward from existing rule-based techniques. ML-RCA is independent of the network technology or topology, and automatically derives the relationship between network element and events without human intervention or predefined rules, which will be key for 5G networks.

Data, data everywhere

Applying these new algorithms to large volumes of data from diverse sources will reveal deep, actionable insights about services and their underlying network resources performance, allowing operators to automate the most complex processes to support self-healing and closed-loop orchestration.

But any given set of analytic tools is never complete without the efficient means to access and view the resulting insights. Analytics-driven user interfaces can unite all operational information, data and tools in a single place. This will allow anyone who needs access to service assurance information to become more efficient, informed and collaborative, allowing them to fully embrace the complex, dynamic nature of new technologies such as 5G.

We foresee plenty of ‘sandbox-style’ testing of machine learning programs

Closing the loop

The goal is what we term ‘closed-loop’ automation, but naturally operators are a little risk averse and reluctant to trust mission-critical infrastructure to automation tools without plenty of evidence of their reliability. We foresee plenty of ‘sandbox-style’ testing of machine learning programs so that the service provider community can reap the benefits of automation already enjoyed by many of the internet scale companies.

The 5G network platform introduces new technologies in the radio, in the cloud core, and in the software defined transport layer, creating a more dynamic networking environment that requires automated, closed-loop, network optimisation to assure SLAs on a service-by-service, network slice-by-network slice basis.

The diverse range of 5G-based service types that are under development – and are crucial to the long-term health and profitability of the service provider community – cannot be delivered on a common infrastructure and technology platform without mastering the art of contextual service assurance.