The Journey to AI-Driven Autonomous Networks

For anyone who attended Mobile World Congress (MWC) in March, talk of artificial intelligence (AI) was everywhere. No matter what booth you visited, AI was the conversation du jour. In some respects, it was a bit overwhelming.

There is certainly no doubt that AI holds the promise of important benefits for telcos. The use of AI and machine learning (ML) across various departments offers the potential to achieve better network optimization and more accurate traffic prediction. This can lead to more effective customer experience management. Leveraging AI and ML also holds the potential for bridging skills gaps and improving operational efficiency.

Beyond incremental revenue and monetization possibilities, these advanced technologies can dramatically enhance early fraud and security threat detection, which opens the door to important cost savings and reputational protection.

But where do you start? Just walking around MWC might leave you with the impression that the next steps will be daunting.

The AI Journey

For communications service providers (CSPs), deciding what to do first in this rapidly transforming environment is the big question. The first step in the AI journey begins with AIOps teams examining how this technology can work in conjunction with existing tools to streamline activities and utilize data from all systems harmoniously. It makes sense to look at all the operations support systems (OSS) tools in the provider’s bag today. Determining how they can work together is the logical place to start.

In the past, network operations teams with deep domain knowledge across multiple disciplines would be tasked with manually sorting through hundreds of key performance indicators (KPIs) and metrics to address trouble tickets and resolve an issue. Given the growing complexity of today’s networks and the vast quantities of data spun off by every network event, business service interaction, and subscriber device, this human-based approach is no longer viable.

Automated workflows based on AI and ML offer a means for gaining insights from the avalanche of data being generated, thus dramatically reducing the need for manual intervention. That said, effective workflow automation will still be dependent on combining customer experience issues with telecommunications network and domain knowledge expertise. The ability of CSPs to strike the right balance will be essential in delivering faster results, reducing weeks of troubleshooting down to a matter of minutes and achieving important cost savings. At the same time, AI- and ML-powered solutions hold the potential to free up engineers to focus on the more intractable issues that don’t immediately lend themselves to automated resolution.

Taking an Incremental Approach to AI

The history of telco tool development has shown steady progress toward the ultimate objective of reducing mean time to resolve (MTTR). As this ongoing journey to MTTR has continued, there have been commensurate improvements in productivity. AI and ML hold the promise of taking this journey to the next level, but CSPs shouldn’t feel as if this is an all-or-nothing proposition. Instead, taking an incremental approach to the adoption of AI can be prudent, allowing CSPs to be flexible enough to shift their strategy if something isn't working as expected.

Key to an effective, incremental AI strategy is equipping teams with AI/ML solutions that deliver efficiency at scale and reduce opex, reshape network services and customer experience, and drive toward experience-driven network construction to enable deterministic experiences.

It’s important to keep in mind that training AI to understand an organization’s network—teaching it how it should behave—takes time. Starting small has its advantages because this allows AIOps teams to properly train AI data models, going through a learning curve much as a human would. Needless to say, this learning curve will be much faster than it is for a human, but it is critically important that the output of the AI be trusted.

Accelerating Your Transformation to AI

The critical first step to gaining predictive insights is to deploy an architecture to collect and process the data at source in order to deliver consistent real-time visibility across the entire network infrastructure. Real-time analysis is far more valuable than interrogating large lakes of historical data that have been dumped offline.

NETSCOUT has been on its own AI transformation journey. Our solutions deliver a stream of Smart Data derived from AI sensors that is jet fuel for AIOps applications, as well as use case–based streams for third parties, such as AWS, Google, Microsoft, Splunk, and ServiceNow.

AI and ML are at the heart of NETSCOUT’s Omnis Analytics solution, providing smart outcome benefits made possible by the best data and deep knowledge of the telco space. These benefits include the move to proactive prediction, the ability to solve “unsolvable” problems, increased accuracy of problem detection, getting at the root of issues faster, increases in optimization, and achieving higher net promoter scores and higher employee satisfaction.

When it comes to AI and AIOps, CSPs are on a journey. The transformation that is taking place holds tremendous promise, but getting there will likely require an incremental approach. Oftentimes, slow and steady wins the race. 
    
To get your network on the first step to AI read more here.