Wake up, Neo: Entering the Matrix behind telecom network operations
- Traditional telecom network management approaches are insufficient for addressing the increasing variability and diversification of service demand that CSPs aim to provide within ‘The Matrix’ of telecom networks.
- The need for a new network operations model arises to enable proactive and predictive network operations through a data-driven and organizational transformation.
- This blog explores how a data-driven model is already paying off and introduces the need for intent-based ops to deal with the growing variability.
The telecom industry has a successful history of managing networks by technology domain, with the basic ideal that if you maximize the availability of the network then you maximize the customer experience. While this inside-out philosophy has served the industry very well in the past decade, it won’t be enough to address the dramatic increase in service demand variability or the augmented diversification outcome communication service providers (CSPs) are aiming to provide.
Arguably, networks nowadays are their own complex matrix. Indeed, modern networks bear a strong resemblance to intricate matrices, akin to the one portrayed in the Wachowskis brothers' 1999 science fiction action masterpiece, The Matrix.
In this film, a simulated virtual reality world is governed by artificial intelligence. Beyond the ethical and moral complexities explored in the storyline, The Matrix left me contemplating two pivotal inquiries: How could the immense volume of data necessary to uphold the matrix be efficiently managed by either humans or machines? And what sort of operating model could sustain such an expansive virtual reality simulation?
In my view, the answer lies in network operations centers (NOCs) - the same type of engine depicted in the film. Twenty-four years after The Matrix premiered, I don't think even the Wachowskis could have imagined the escalating complexity of the NOCs inside today's telecom networks.
In 2018, my fascination with this area led me to eagerly seize the opportunity to spearhead a groundbreaking project aimed at revolutionizing Ericsson's network operations model. We started with a blank sheet of paper, aiming to define a new operational approach that would shift CSPs from reactive and incident-driven operations to predictive and data-driven operations.
As part of their individual journeys towards more predictive and data-driven operations, we realized that CSPs needed to transform in two essential dimensions: technological and organizational.
Reimaging this technological transformation entails a fundamental overhaul, with CSPs needing to move towards an operations environment that is orchestrated end-to-end in an intent-driven fashion.
Meanwhile, the organizational dimension necessitates a cultural transformation, re-engineered processes, new ways of working and the development of skills that CSPs need to adopt internally in order to fully become a data-driven organization.
During our research, it became clear that without addressing the organizational dimension and undergoing proper operational transformation, CSPs could run the risk of merely adding complexity with new technology, leading to increased costs.
While transforming operations has been a constant state for most CSPs, the pace of change has been unprecedented. Most CSPs have been transforming too slowly, and ongoing effort is required to seize new opportunities such as those at the network edge.
The engine behind The Matrix
The culmination of our project was the launch of the Ericsson Operations Engine (EOE) in 2019, a multi-technology and multi-vendor data-driven approach to operating and optimizing telecom networks. EOE serves as a catalyst for CSPs to thrive in today’s dynamic landscape of 5G networks, providing an agile and holistic solution that intelligently facilitates a seamless transformation. EOE capitalizes on people and domain expertise, data- driven processes, modernized applications and platforms, as well as automated and AI-driven insights.
Today, I can proudly say we have succeeded. Through global feedback loops, we have achieved an unapparelled scale that allows us to handle a Matrix comprising more than 710 thousand sites, 18.2 M work orders and 6.4 M RAN cells with highly focused human intervention. This feedback looks like it will continuously evolve our processes and expand our libraries of closed-loop automation and AI use cases. To illustrate, today the aggregated level of closed-loop automation within our network operations exceeds 88 percent and in 2022 alone, we reached 36 million AI recommendations.
This has a significant tangible impact on CSPs’ operational business outcomes. From a network performance perspective, data-driven operations reduce network unavailability by 34 percent, while also decreasing customer complaints by 21 percent.
The results speak for themselves. The so-called “Triple As” – analytics, automation and artificial intelligence – have been, and will continue to be, a key element in this story. However, if we want to continue leading the CSPs’ transformation journey we, can rapidly respond to CSPs’ needs with minimal manual intervention by translating commands (intents) into network-based actions.
Certainly, intent-based operations are an inevitable shift in the approach to monetizing 5G investments. Network will need to support a dramatic increase in the variability of service demand, and the addition of more humans to cope with this demand will not be scalable.
We have identified three main drivers for intent in building and operating new services:
- The complexity and cost of operating 5G networks and beyond require a new level of automation, surpassing even the capabilities of current AI/ML automation.
- There is a need to transform operations to better align with the needs of the business.
- Transparency and AI explain ability are crucial when it comes to tracing back all the decisions recommended and acted upon by the system. This level of visibility allows CSPs to increase their trust in AI and ML technologies. Obviously, our EOE operating model is evolving to support a self-optimizing network driven by intent and underpinned by hyper-automation. This transition involves shifting from people managing the network to people managing the machines that oversee the network. For this approach to work, data/knowledge, policy, automation, assurance, analytics, machine learning and reasoning and security must be integrated into a true intent-based network, resembling the Matrix.
The next evolution of data-driven operating models
The next evolution of data-driven operating models will need to support a wider variety of applications and use cases for consumers and businesses. CSPs must aim to provide configurable services with detailed agreements on functional and non-functional characteristics that require dynamic adaptation to the network. This calls for a sort of hyper-automated “Matrix reloaded” scenario, where telecom networks must be managed in real-time to deliver these requirements. Concepts like "autonomous networks" and "intent" will be essential for driving new propositions and automating´ the network's state to meet performance key performance indicators service level agreements and business outcomes.
During FutureNet World event earlier this month, I had the opportunity to meet industry analysts and customers and gain a firsthand view from various sessions. I heard directly from CSPs about how they are anticipating a radically different future for their companies, particularly in terms of repositioning their business as solution providers in response to the demands of enterprises from telcos. This need was really well encapsulated in the Enterprise Opportunity presentation, which was delivered by STL partners.
There were also lots of interesting talks around Open Telco and Networks as a Service (NaaS), in particular the transformation challenges associated with a disaggregation strategy was a hot area of discussion. That being said, it was clear that addressing all the challenges ahead will require robust data-driven operating models that, as highlighted during my presentation, are already paying off for CSPs.
Watch on demand my presentation @FutureNet World 2023.
And finally, revisiting the Matrix analogy, although we have somewhat ventured into the world of the Matrix trilogy, However, on a positive note, we have gained the knowledge and expertise to effectively "manage" the network.
Oct 14, 2020 |Blog post
Automation, AI and machine learning, Cloud infrastructure
Sep 04, 2023
Sep 04, 2023
Like what you’re reading? Please sign up for email updates on your favorite topics.Subscribe now
At the Ericsson Blog, we provide insight to make complex ideas on technology, innovation and business simple.