People using their mobile devices at a concert.

Autonomous networks

Experience networks that adapt to your business objectives in real time

Available in:

English
简体中文

Autonomous networks mark a new era for networks – replacing traditional rule-based automation with autonomous operations capable of thinking, learning, adapting and making business-critical decisions in real time with zero human intervention.

Autonomous networks are built on the key pillars of agentic closed-loop automation, intent-driven interactivity and interoperability, intelligent data management and open platforms and APIs. At the very foundation sit the network’s autonomous domains, each equipped with self-x capabilities that enable domains to deploy, configure, monitor, maintain and retire by themselves.

Through autonomous network operations, the dots are connected. Dynamic intent management and agentic AI closed-loop automation ensure that high-level business objectives are translated into actionable network updates: analyzing intents, adapting to changing conditions and resolving conflicts to orchestrate and maintain the delivery of advanced services across all OSS/BSS layers, vendors and network domains.

In doing so, autonomous networks lay a crucial foundation for new business paradigms, making it possible for networks to streamline, scale and shorten the time to market for advanced network services – from differentiated connectivity to APIs and beyond.

Intent-driven networking provides a key step on the journey to truly autonomous networks. As leaders in this space, we’re ready to guide our customers through this transformation.

Erik Ekudden, Ericsson CTO

Autonomous networks give communications service providers (CSPs) and enterprises an AI ready, intent driven foundation that can support differentiated connectivity and demanding networks for AI use cases at scale. By combining AI, closed loop automation and autonomous domains, they help turn complex 5G and future 6G capabilities into predictable business outcomes that are easier to launch, assure and monetize.

Business and service agility

Autonomous networks dramatically shorten time‑to‑market by translating high‑level business intents into end‑to‑end services, so differentiated connectivity and AI application offers can be introduced, adapted and retired at the pace of demand. This enables SLA‑backed services and use case exploration without re‑engineering the network each time.

Network performance and resilience

Autonomous networks keep performance predictable, as demanded by AI applications and mission‑critical traffic by continuously aligning resources with service intents and SLAs, rather than relying on best‑effort behavior. Issues are anticipated and mitigated before users are impacted, safeguarding consistent quality for high throughput and latency‑sensitive use cases.

Simplification and efficiency

A clear separation of concerns across autonomous domains means each domain optimizes within its scope while the overall system maintains end‑to‑end intent fulfillment. Operators set goals and guardrails instead of managing thousands of low‑level tasks, reducing operational effort, improving energy efficiency and lowering total cost as AI adoption scales.

Translating business intents into network actions harnesses the network’s ability to think, reason and adapt with cognitive autonomy. This is performed through numerous autonomous network agents that work together to analyze the network state, identify anomalies and causes, search and test different solutions and execute corrective actions in real time. Throughout the process, agentic AI continuously learns from outcomes to improve future responses.

Define

Business intents are expressed, requested and confirmed through an intent management loop that takes place across each autonomous domain. New services are proposed, evaluated and requested (via the orchestration function) based on their suitability to satisfy the intent. Through a separate closed cognitive loop, the process of intent assurance then begins.

Observe

Autonomous network agents are put to work to interpret intents and assess the current network state. This takes place through an assurance agent that collects, analyses and leverages network data to monitor and assess the health of the newly activated service. If the intent is not met, a closed loop activity is triggered.

Decide

When a closed loop activity is triggered, utility-based evaluation helps the system determine the most valuable course of action, even in the face of conflicting requirements or new conditions. This allows for dynamic prioritization and decision-making aligned with business goals.

Act

Through an actuation agent, the network takes corrective action – generating the required modification requests through the orchestration function. Based on current resources and topology, the service(s) is re-designed, calculating the appropriate execution plan that will be used to alter the resource configurations in the network.

Realizing autonomous networks: from talk to accelerating implementations

Executives from AT&T, Bharti Airtel, Swisscom, Telstra and Ericsson discussed the new autonomous network landscape, including how to unlock the full value of network autonomy with the help of AI and intent.

Find out more

Prathet Tankuranun, CTO, True Corporation.

True: Automation is at the heart of transformation

True Corporation CTO Prathet Tankuranun discusses how automation and partnerships are at the core of his network and operational transformations.

Read now

Explore key technology enablers

A major step towards autonomous networks is the shift to intent-driven networking, where high-level requirements guide network behavior and service assurance. Together with AI-driven data management and agentic closed-loop automation, it enables greater network autonomy, accelerated by Open APIs and standard interfaces across autonomous domains.

Autonomous systems require an AI-driven framework that supports service operation, integration, creation and assurance processes in a non-disruptive manner at run time. Talk to your network is an umbrella term that brings together key AI technologies and enablers to establish autonomous networks capable of addressing those areas.

The talk to your network framework integrates knowledge and agent-driven methods to transform natural language into actionable intelligence. Talk to your network completely changes the way you can interact with OSS/BSS and the network domains to drive different types of tasks and processes, as it leverages, as it leverages three AI technology pillars of autonomous networks: generative AI, agentic AI and intent-based AI. Together, these pillars contribute to an integrated system capable of conversation, orchestration, intent decomposition, and autonomous execution.

Generative AI

Generative AI uses machine learning (ML) algorithms to create new content, including text, images, audio, or code. Large Language Models (LLMs) have reached human-level quality in generating text and answering questions. In telecom, this technology can be applied to both customer-facing solutions and advanced network solutions, such as conversational AI-based assistants.

Agentic AI

Agentic AI refers to AI systems that can sense, think, adapt and act using planning algorithms and adaptive decision-making. Such systems can be implemented as a single agent or a coordinated multi-agent architecture where multiple agents collaborate towards defined goals without human guidance.

Intent-driven AI

Orchestrates agentic AI entities, aligning their actions with overarching service intents and business goals for seamless coordination and execution.

Reinforcement Learning

Reinforcement learning refers to AI systems that learn how to make better decisions over time by interacting with their environment and optimizing for long term outcomes. These systems enable agents to adapt autonomously in dynamic network conditions, enabling networks to self optimize and deliver more intelligent, zero touch operations.

Telecom foundation world models—Toward higher levels of network autonomy

Adapting Agentic AI for intent-driven, closed-loop autonomy

Assessing the impact of cognitive rApps in achieving autonomous networks Level 4

How rApps Power Autonomous Network Evolution - Ericsson

Autonomous network operations: from reactive management to intent-driven optimization

AI agents in telecom network architecture - Ericsson

An autonomous network is subdivided into autonomous domains according to the generic architecture proposed by TM Forum. Autonomous domains are allocated within layers of the network operation software stack, residing on the business, service or resource layer. Within a layer, multiple autonomous domains may exist in parallel or in further subordinate layers, each with terminal goals that it is required to meet and optimize. To reach its goals, it also needs other domains to contribute to and comply with certain requirements.

Intent is used to communicate and manage these requirements in the various layers in our layered approach to intent management. An intent manager plays the role of intent owner if it defines requirements with which another domain needs to comply. If, on the other hand, it receives requirements that it must comply with in its own domain, it assumes the role of intent handler.

The role of intent at each layer

At the business layer, intent captures customer expectations and business goals, replacing traditional service level objectives (SLOs) with dynamic, goal-driven requirements. These intents are defined during service ordering and negotiated through APIs, then passed to the service layer for fulfillment. Here, the focus is on intent-based assurance that monitors service performance, predicts SLA violations and takes corrective actions based on business value.

At the service layer, business intents are translated into actionable service and resource requirements. Here, the service orchestrator dynamically selects and configures services based on intent. The system continuously monitors compliance and only acts when requirements are not met. This layer decouples service logic from specific resources, allowing more flexible and AI-driven orchestration. It enables real-time autonomous healing, optimization and service adaptation, ensuring that services evolve with changing conditions and demands.

In the network domains in the resource layer, intent is handled at two levels: the management layer and the network function layer. For example, the RAN domain’s intent manager ensures compliance with service-layer requirements using analytics, orchestration and AI agents. If deemed necessary, existing service instances can be reshaped to use new or more efficient resources. Intent-aware RAN functions monitor real-time performance and adjust configurations or control loops to meet evolving requirements. This and similar implementations for the other network domains enable the network to self-optimize and evolve based on overarching business objectives.

Embarking on a multi-year autonomous network journey requires a holistic transformation where data management ensures trusted output, AI drives intelligent automation, cloud and IT provides scalable infrastructure, service orchestration enables agile delivery, and monetization strategies convert technological advances into revenue.

However, no one journey is the same. With different tech environments, business priorities and deployment challenges to consider, each transformation journey must be designed and paced according to each unique context. A strategic framework provides the foundation for this, meaning that key transformation areas can be identified, measured and prioritized systematically. This also provides a basis to benchmark progress and address gaps that can accelerate the journey to full autonomy.

Basic automation supports human operators through scripted tasks, with limited data integration and manual deployment of monolithic OSS/BSS systems on traditional hardware. Service orchestration is slow with high error rates, relying on manual processes and traditional interfaces. Offerings are based on basic service tiers differentiated by bandwidth and price points.

Workflow-based automation is deployed across specific domains using static rules supported by centralized data repositories. Virtualized infrastructures and predictive AI/ML models lay the basis for some basic automation, however human oversight remains essential for critical decisions. At this level, services become more dynamic, featuring policy-driven connectivity, event-based charging and early-stage integration with OSS for usage data.

AI/ML models begin to support decision-making for discrete tasks, enabling policy-driven automation with reduced human intervention. Real-time data pipelines and cloud-native containerized microservices allow systems to detect and respond to events dynamically. Cross-domain orchestration platforms are introduced to improve visibility and enable closed-loop automation for reactive self-management and auto-remediation. Services are created, deployed and managed based on AI-driven insights, with near real-time OSS integration for usage data.

All tasks are automated for certain processes based on declarative goals (intents) with human supervision supported by AI/ML driving decision-making, predictive analysis, and real-time adaptation across domains. Cloud and IT environments are self-optimizing and leverage reinforcement learning and predictive scaling. Closed-loop automation enables continuous performance tuning, while differentiated connectivity services adapt to network and application needs. OSS feedback supports real-time assurance and monetization orchestration.

All operations are fully automated with no or minimal human intervention. A unified data platform enables end-to-end governance, self-service analytics and AI-driven automation. Cognitive systems continuously learn, adapt to unforeseen challenges and optimize long-term outcomes using advanced techniques like causal inference and anomaly detection. Cloud-native IT environments are entirely self-managing and self-healing. Cognitive orchestration autonomously translates business goals into actions, while a self-learning monetization engine dynamically adjusts offers, pricing and bundles based on real-time insights into customer behavior and network conditions.

Explore our OSS/BSS e-brief to find out what’s needed to advance your journey.

Aiming to achieve Level 4 autonomous core networks? Discover the key drivers, challenges, and strategic priorities influencing service providers transition to autonomous core networks.

This study from Omdia maps service providers transformation journeys toward autonomous core networks. The report offers practical guidance on operating-model changes and the skills and roles required.

Read study

Embarking on the journey toward autonomous networks requires a carefully crafted strategic roadmap that harmonizes technological advancements and investments with your overarching business goals.

By setting a strategic direction, you can effectively navigate the complexities of transitioning to autonomous networks, ensuring that your technological investments are aligned with and drive your business goals. This balanced approach not only facilitates innovation but also guarantees that your network evolution is sustainable and value-driven.

Deconstruct new services

Transforming business outcomes into precise functional requirements is crucial for transitioning from "best effort" services to guaranteeing premium Service Level Agreements (SLAs). This involves:

Understanding business goals: Clearly define what success looks like for your organization and how autonomous networks can support those objectives.
Mapping requirements: Break down these goals into specific technical and operational requirements that the network must fulfill to deliver on premium SLAs.
Iterative refinement: Continually refine these requirements as your business and technological landscape evolves.

Choose the right path

Selecting the optimal route towards autonomous networks requires alignment with your unique business objectives. Considerations include:

Business alignment: Align network capabilities with strategic business priorities, such as improving customer experience or reducing operational costs.
Scalability and flexibility: Choose a path that allows for scalability and adaptability to future technological advancements.
Risk management: Evaluate and mitigate potential risks associated with the transition to higher levels of network autonomy.

Invest for value

Developing an investment strategy for 5G autonomous networks requires a guiding framework that ensures alignment with long-term goals. Key elements include:

Strategic prioritization: Prioritize investments based on potential ROI and alignment with business objectives.
Technology assessment: Continuously evaluate emerging technologies and their potential impact on your network strategy.
Value realization: Focus on realizing tangible benefits from investments through improved network performance and new service offerings.

A common vision for autonomous networks

Today, Ericsson plays a decisive role in shaping a common industry vision for autonomous networks across standardization fora such as the TM Forum, 3GPP and Open RAN.

TM Forum provides a foundational reference architecture and intent-management APIs, while 3GPP extends these concepts for specific services. O-RAN is working to unify these approaches to support an open, intelligent RAN.

Alignment of these activities across standards bodies and open-source communities will remain crucial to building an open, interoperable ecosystem for autonomous networks.

Autonomous networks

The basics explained

Benefits of autonomous networks

How it works: from intent management to closed-loop assurance

Follow the journey to autonomous networks