Achieving 5G Service Assurance Through Automation
Operators need innovative techniques to achieve visibility into mobile and fixed networks.
As the introduction of 5G service adds greater complexity to today’s multitechnology, multigeneration, and multivendor networks, the ability to leverage next-gen, intelligent Automation techniques to proactively achieve service assurance is more important than ever.
Operators implementing 5G must grapple with disaggregation, cloud-native technologies, and other technologies such as Voice over New Radio (VoNR) and edge computing. As a range of new services are introduced, operators face significant changes to network management, particularly network operations, resulting in greater demand for service assurance. End-to-end network service visibility and automation will be needed for security and assurance across infrastructure and applications.
Fortunately, a new generation of solutions is available to help deal with this growing complexity and demand for greater operational flexibility. Automation is key to managing new 5G infrastructures and the diverse services that flow from them.
The Need for Automation
When it comes to managing new 5G infrastructures, network automation is vital for addressing the need to simplify operations as networks scale and achieve increased operational efficiency. Automation is also crucial for improving resource utilization—responding in real time to conditions as the network changes and making adjustments. The ability to do more with fewer people is particularly important during a time when skill shortages are pervasive.
Automation will play a key role in helping to overcome the complexity of 5G service assurance as well as of the entire network, including critical aspects of the 4G network. Over-the-top (OTT) services, regardless of whether they are on 5G user planes or fixed networks, will benefit from intelligent automation solutions.
When performing end-to-end service assurance, several high-level concerns tend to be top of mind for network professionals. When issues arise, are they related to the radio access network? Or are they related to the 5G core signaling? Is latency being introduced within the hyperscaler? Is latency within the service level agreement (SLA) thresholds for the ultra-latency slice? Are subscribers experiencing coverage issues, and if so, are the issues in a specific market? Could problems be related to back-end data services, such as slow DNS response? Determining the root cause of such issues in highly complex networks is no small undertaking.
The Challenges of Supporting a 5G Standalone Network
Being able to support a fully 5G standalone network is one of the largest, most complex transitions carriers will undertake. This migration will be a giant leap when compared with going from 3G to 4G. Beyond the radio access nodes that started to change with 5G non-standalone, 5G standalone comes with a new set of core network functions. With those core network functions, we also see the introduction of new message formats, along with the option of encryption on the service-based interface, which creates further challenges.
As the transition from dedicated appliances to highly complex virtualized cloud-native architectures speeds up, accessing 3G packets between new network functions is becoming far more complex. For some carriers, the 4G and 5G nodes are collapsing, offering a single, dual-functional node. What this means for service assurance is that carriers will be compelled to monitor the 4G network, while at the same time monitoring the 5G standalone network in order to perform any type of triage as the migration to VoNR takes place. With 6G on the horizon, these challenges will only become more complex.
The notion that large amounts of data will be aggregated into an extensive data lake where data scientists will run artificial intelligence (AI) and machine learning (ML) algorithms to quickly and easily identify issues is something of a pipe dream. Science teams end up with too much raw data and generally lack proper telecom domain knowledge. And applying AI/ML techniques without the proper insertion of domain logic in the right place ultimately leads to data bloat, while still failing to deliver needed results.
Engineering and operations teams may have day-to-day hands-on knowledge of the problems, but what they truly need is a way to operationalize domain knowledge into an automated system. To avoid the typical high rate of false positives with traditional AI/ML approaches, what is needed is an ecosystem that will allow automation with a common language for quickly blending 5G standalone network packet level data—what NETCOUT calls Smart Data, domain knowledge, and ML techniques to solve complex problems. By combining data science techniques that are focused on data reduction, teams can keep only what is necessary for deriving use cases.
Making the Case for an Intelligent Automation Framework
To be able to quickly identify and pinpoint the source of issues, such as call drops, engineers need a 360-degree view of all aspects of the network. It is hypercritical to be able to examine not only the core side of the network but also the radio access side. And if services are coming from the multi-access edge computing (MEC) architecture, then engineers will need visibility here as well to ensure issues are not cropping up inside of the hyperscaler architecture.
Finding the root cause of issues typically takes hours, if not days or even weeks. However, with intelligent automation, troubleshooting can be consolidated to minutes across all markets, calls, and customers—7/24, 365 days of the year.
True intelligent automation requires a framework that leverages a superior detection approach that relies on multiple AI/ML algorithms in parallel and efficient use of domain knowledge blended with AI/ML to get at the root cause of issues. Such an approach will be capable of discovering hidden patterns, as well as discerning signal versus noise and automatically determining the first point of failure. Intelligent automation will be able to detect customer-impacting versus nonimpacting failures. It will offer end-to-end monitoring of the health of a service, from RAN to control/user plane and mobility and isolate issues to RAN versus the core.
An intelligent automation framework that leverages packet level data, subject-matter-expert (SME) domain knowledge, and adaptive AI and ML technology can drive outcomes with the most business-relevant impact on the network. To achieve service assurance, the right automation framework must be put in place. That framework should deliver end-to-end visibility, examining the RAN, the core, and the MEC architecture, as well as LTE, Voice over LTE (VoLTE), and 5G standalone or 5G enterprise. Such an architecture needs to be pliable, allowing carriers to define basic business models. It needs to be cloud native so it can scale up and down along with the network, which is especially important in a 5G standalone network. It also needs to have the capability to send information that can be seen visually by a human, but also viewed by a mitigation system or for closed-loop orchestration.
An intelligent automation framework can provide an environment that’s pliable enough and adaptable enough to not only solve highly complex service assurance issues facing 5G standalone but also to help manage OTT issues, as well as be used for RAN to tackle security issues cropping up today.
Watch the webinar.