30 APRIL 2020

By: Dima Alkin, VP Service Assurance Solutions, TEOCO

5G is set to bring exciting new use cases to the table, but with it, it’ll also bring heightened complexity. And while degradations in quality of service (QoS) or quality of experience (QoE) may result in frustrated subscribers, when it comes to enterprises, just an hour of network downtime can cause a significant loss in revenue, amounting to millions of dollars. What’s more, for the mission-critical applications that will be enabled by 5G, network downtime could be the difference between life or death, which is quickly becoming more than just a scary statement as made evident by current events.

The reality is that CSPs cannot continue to monitor their network in the same way they have done with 4G and previous generations of mobile technology—in fact, it’s not that they can’t, but they simply won’t be able to. A 5G network will see the number of data sources—from network performance and telemetry, to events and alarms—increase on a dramatic scale requiring CSPs to think of a new way of monitoring their network infrastructure. Many aspects of this new way will come in the shape of automation enabled by AI and machine learning.

Proactive not reactive
Traditionally, CSPs have taken a reactive approach to monitoring and assuring their network assets—that is, addressing faults or outages once they have already occurred, or are occurring. While in a 3G and 4G world, this reactive approach was enough—or at least, manageable, that won’t be the case with 5G. Instead 5G will require a much more proactive approach to network performance management. This will be particularly important as CSPs seek to make money from the enterprise market.

To allow CSPs to monetize this opportunity, they will need to be able to comply with strict contract-backed SLAs that guarantee a certain level of service. These SLAs will be critical for the enterprises reliant on connectivity to ensure they can offer their customers the right experience and service—for example, eGaming companies who will need to ensure that players can take part in online game sessions without the risk of latency getting in the way; or perhaps augmented reality application providers that rely on ultra-reliable, low latency communications (URRLC). These enterprises need to know that their users or customers get what they’ve paid for. If 5G enterprise SLAs aren’t met, not only will this impact CSP enterprise customers, but it will also come with significant financial penalties for CSPs—a cost they simply cannot afford, given all the investments they are making in building 5G network infrastructure.

This is where this proactive approach comes in. By being able to anticipate and predict end-to-end service availability and prevent network performance issues before they occur and impact the end-user, CSPs will avoid the risk of incurring financial penalties, while making their operations more efficient and, rather importantly, while maintaining a good reputation among their enterprise customers. So, what does this proactive approach really look like?

Proactivity and automation
When thinking about network performance management, we can’t think about proactivity without automation. In a 5G world, there will be more data sources—network data, and customer usage records of all kinds, telemetry etc. —than ever before. While this potentially brings CSPs greater insight and context into what is occurring across their network, it also presents some challenges, namely: how can humans alone analyze and understand all these different data points to identify network performance issues or abnormal network behavior? The short answer is they can’t.

Instead, CSPs will need to bring in automation capabilities by leveraging AI and machine learning. AI and machine learning can sift through numerous data sources to quickly identify or repair network failures or issues. Doing so benefits CSPs in a number of ways. First, it allows them to automatically detect network patterns as they occur and arise. Second, it enables the automatic and proactive filtering and prioritization of alarms. Third, it gives CSPs the ability to instantaneously identify the root cause of network outages, performance degradation and congestion issues. In addition, automation can also help CSPs understand evolving network trends and anomalies, and predict how said events may impact services or subscribers. This level of automation gives CSPs unprecedented insight and visibility into their network and its behavior. In addition, the ability to be able to detect non-trivial issues at an early stage, or even before they occur, gives CSPs the opportunity to take preventative measures to minimize the impact of said issues.

Automation will also play an important role in helping CSPs to manage individual network slices. Each network slice will have its own characteristics and requirements in accordance to the service or customer it serves. For example, a remote healthcare application may use one slice with its own requirements, while an eGaming company may use another, also with its own requirements. Thanks to network slicing, neither slices—or events occurring on either slice—impact the other. Managing and monitoring these slices is no easy feat, and so CSPs will need to lean on automation to be able to collect, process and analyze large volumes of data across all network layers, and correlate these with different network functions and applications.

Enterprise 5G presents a new and exciting opportunity for CSPs—it represents their golden ticket to monetizing and achieving 5G ROI. But doing so will require new thinking and new approaches. The assurance and network performance management systems of yesterday, simply won’t work for the networks of tomorrow. More data and greater complexity mean network engineers alone cannot do it all, and so in a 5G enterprise world, we’ll see machine learning, AI and automation become critical assets for CSPs wanting to maintain and guarantee SLA-grade connectivity.

Originally published on 30 April 2020 in Vanilla Plus