5G is here. What do you need to know about intent-based networks?
Intent-based products such as Siri, Alexa and Google Assistant have redefined our daily lives. But could they soon redefine our networks also? Find out what the future holds for intent-based networks below…
It’s a well-known fact that future networks (5G and beyond) will be highly complex. With billions of connected devices, sensors, smartphones, and industrial and massive IoT connections, these networks will be difficult for service providers to manage.
Without automation, AI and analytics in network operations, it’s impossible for the human brain to comprehend the complexities involved in managing such networks, or understand how to provide a great customer experience. With an increasing trend towards customer centricity, operators are undertaking major digital transformation programs that embrace automation and newer technologies. These are essential for managing complexity, reducing OPEX and transforming from traditional telecom operators into the digital service providers of the future.
The first wave of digital transformations marked the beginning of a long and complex journey towards zero-touch networks and service management. However, there is still more to be done. This blog will focus on how future networks will evolve and top-level business intents will be achieved with minimal human intervention.
What are intent-based networks?
The networks of the future will be intent-driven, meaning a business intent will be defined and the network will adapt to achieve the desired results – free from human intervention. The journey from defining intents to achieving them may involve multiple iterations and learnings from previous steps. While the ideas behind intent-based networking have been discussed for several years, the concept has picked up pace due to advancements in machine learning, automation, analytics and orchestration.
There are already many examples of intent-based products in our everyday life. Think of virtual assistants like Siri, Alexa or Google Assistant, who hear our commands and deliver what we ask for. Google maps and other navigational services are all intent-based, with users selecting the destination and leaving it up to the application to find the best route for the journey.
A business intent in telecom networks could be, for example, configuration of a particular network topology, or configuring few 5G NR cells in a city’s crowded downtown area. More complex intents could be finding sleeping cells in a network for the next 7 days or improving the customer satisfaction score by 2 percent points in a cluster of cells. This top-level requirement is then broken down into multiple smaller intents, fitting each of these requirements into layers – with each layer then striving to achieve its own intent. These different layers of the intent model don’t need to achieve the top-level requirement, they focus on achieving smaller, local intents which coincide with specific service level agreements.
The zero-touch ambition and the building blocks of intent-based networks
As we strive toward zero touch networks an “intent-based everything” approach will be key. The two most important building blocks for an intent-driven network are:
- Full closed-loop automation: with this, there will be no human involvement in the automation loop. This will be guided by a real-time, intent-based feedback loop of communication whereby networks monitor themselves and automatically fix problems with the help of AI frameworks. From the time the business intent is set, it is important to achieve the goal in a specified timeframe by selecting the right policies and translating business intent into policies at every step, including IT infrastructure, RAN/core nodes, orchestration and applying learnings from AI frameworks
- Service assurance with knowledge feedback: achieving an intent can be an iterative process and may involve multiple learning and automation feedback loops. With the rise of analytics and machine learning modeling techniques, it is possible to verify if the desired intent is applied and if business outcomes have been achieved or not. Machine learning systems work on finding patterns in large data sets. If there are few or no structured inputs to find patterns, machine learning systems can’t solve a new problem – as there will be no apparent relation to its prior knowledge. The limitations posed by machine learning modeling are solved by applying machine reasoning to our systems. Machine reasoning focuses on understanding relationships among data and deducing new information. These systems will utilize and combine the information from various data sources, automation loops, and domain experts into a knowledge source which can be used in the next iteration to achieve the intent.
In fact, the motivation towards intent-based network stems from the Software Defined Networking (SDN) architectures. With SDN the process of network management was automated, but automated processes that translate business intent to policies and ensure the network meets the requested goal were still missing. SDN remains a building block in the migration towards intent-based networks.
How do intent-driven networks work?
The figure below highlights how a top-level business intent is achieved using policy frameworks, orchestration and analytics alongside each other in a fully closed automation loop. The top-level business intent is mapped to meet intermediate network/node level requirements and broken into goals which closed-loop automation can address. These goals can be small enough for a single closed-loop automation cycle to address them, or big enough to require multiple iterations and learning loops.
In the scenario above, the operator wants to transform its business by focusing on providing a best-in-class customer experience to its subscribers. The top-level business intent is broken into network goals which could be related to service KPIs and a service level index (SLI). The service KPIs are then calculated and developed based on data from network nodes, probes, OSS/billing data etc. The reasoning loop will continue to break the intent into smaller goals until the top-level goal is reached.
There are plenty of other examples which could be realized in the coming years thanks to advancements in automation, orchestration, and analytics frameworks. Here are some other use cases for service operators:
- An operator wants to improve profitability and sustainability
- An operator wants to reduce their carbon footprint, for example, by identifying its most energy intensive sites and making plans to reduce energy consumption
- Reducing OPEX, for example, by keeping the number of radio tower inspection visits to minimum
- Moving a legacy billing system to distributed cloud
- Dynamically creating, managing and deleting network slices from orchestration
Networks in 2025 and beyond
In five years, our networks will be much more complex. They’ll handle a huge amount of data, and IoT will create an explosion in the number of end devices. Intent-driven networks with intelligent machine reasoning and full closed-loop automation provide the best opportunity for operators to create the automated environment that their businesses will require in future. We’re still a few years away from fully intent-based networks, but we’re getting closer. The benefits of such networks will be seen when an intent-based approach is deployed across all network domains.
These future intent-driven networks are part of a bigger technology shift in telecom, where telco expertise will be combined with AI. At Ericsson, our AI solutions and services are integrated into a network built by experts to address the challenges of service providers.
At Ericsson we call it “AI by Design”.
In our operations engine podcast series, we hear what industry experts and leading engineers believe will happen next in the rapidly-changing environment of network operations.
Visit our managed services page to read more insights about the next generation of networks.
Read more about the zero-touch customer experience.
Visit Ericsson Research’s autonomous networks page to learn how the future of networking looks.