Samsung has detailed how autonomous network architectures and agentic AI are replacing manual telecom operations.

Mobile network operators face pressure from shifting traffic patterns between business centres and residential zones, as well as temporary spikes during large public gatherings. Conventional operational models require engineers to adjust capacity manually, introducing latency and service degradation.

Autonomous systems help to address these issues by using predictive machine learning to detect and resolve traffic surges prior to subscriber disruption. According to the TM Forum framework, autonomous networking spans two dimensions: Self-X operational capabilities across the network lifecycle and Zero-X (defined as zero-wait, zero-touch, and zero-hassle for operators and users) outcomes.

Autonomous operations divide across three operational layers – Business, Service, and Resource – and are connected via closed control loops across independent operational domains. Operators issue high-level business goals through intent-driven interfaces, which the lower layers translate into execution commands without continuous human direction.

Six maturity levels define deployment progression

The industry measures autonomous capability across six levels, extending from manual operations at Level 0 to complete closed-loop management across every operational domain at Level 5.

In a TM Forum survey of 80 global operators, 20 percent reported plans to deploy Level 4 systems by 2027, with 81 percent targeting Level 4 or above by 2030.

Reaching Level 4 and Level 5 requires a five-stage cognitive loop:

• Intent: Operators define high-level parameters, such as reducing power usage while maintaining radio coverage.
• Awareness: Observability tools ingest logs, multi-dimensional telemetry, traces, and topological states.
• Analysis: Machine learning evaluates discrepancies between network performance and intent targets, identifying sleeping cells and traffic loads.
• Decision: Systems simulate and select resolution models that prevent service-level agreement conflicts.
• Execution: Standardised open APIs run automated configuration and repair actions across multi-vendor infrastructure.

Telco-grade models and multi-agent coordination

Specialised large language models provide domain-specific reasoning for operational workflows. Beyond automated anomaly detection in radio access networks, telco-grade models evaluate urban planning data to guide cell-site capacity expansion and trigger automated cybersecurity countermeasures.

These systems also allow operators to package network performance into commercial APIs. Enterprise customers can purchase guaranteed quality of service and deterministic latency for mission-critical traffic profiles.

Agentic systems represent a structural departure from monolithic automation. From Level 0 to Level 3, algorithms serve primarily an analytical function by presenting recommendations for human sign-off.

At Level 4, independent AI agents execute actions directly based on high-level operational parameters. Level 5 systems remove external parameters entirely, allowing agents to generate operational objectives directly from telemetry insights.

Samsung is deploying this model through its O-RAN compliant CognitiV Network Operations Suite (NOS). The platform runs an AI Agent Fabric combining generative models with domain software. The architecture uses Retrieval-Augmented Generation (RAG), Model Context Protocol (MCP), knowledge ontologies, cross-domain datasets, and digital twins to manage planning, rollout, maintenance, fault isolation, and optimisation.

Standards development remains distributed across multiple bodies. TM Forum oversees Open Digital Architecture specifications, the O-RAN Alliance develops open radio interfaces, 3GPP manages global technical specifications, and the AI-RAN Alliance explores operational deployment models.

Samsung currently holds leadership posts across 3GPP Technical Specification Groups, co-chairs working groups within the O-RAN Alliance, and leads the Agentic AI Task Group as Vice Chair of the AI-RAN Alliance Board of Directors.

See also: Ericsson Japan begins AI network research with Tokyo partners

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including the IoT Tech Expo and Cyber Security & Cloud Expo. Click here for more information.

Telecoms is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.