Inferensys

Glossary

RAN Function Exposure

The capability of the E2 interface to abstract and expose specific RAN control and data collection services to xApps via a standardized API, enabling vendor-agnostic optimization.
Data scientist building training data pipeline on laptop, data preprocessing visible, technical workspace.
E2 INTERFACE ABSTRACTION

What is RAN Function Exposure?

RAN Function Exposure is the capability of the E2 interface to abstract and expose specific RAN control and data collection services to xApps via a standardized API, enabling vendor-agnostic optimization.

RAN Function Exposure is the architectural mechanism defined by O-RAN that abstracts vendor-specific radio resource management algorithms into standardized, discoverable services. Through the E2 interface, the Near-RT RIC queries the E2 Node to retrieve a list of supported RAN Functions, each identified by a unique ID and defined by a formal service model. This abstraction layer decouples the optimization logic in xApps from the proprietary implementation of the underlying hardware, enabling a multi-vendor ecosystem where a single xApp can control radio resources across different equipment suppliers.

Each exposed RAN Function is described by an E2 Service Model (E2SM), a standardized ASN.1-based specification that defines the event triggers, actions, and reporting structures available. Common exposed functions include KPI monitoring, UE admission control, and bearer context management. By exposing only the granular control points necessary for optimization—rather than the entire internal logic of the gNB—the architecture preserves vendor intellectual property while granting the RAN Intelligent Controller the fine-grained telemetry and actuation points required for AI-driven closed-loop automation.

E2 INTERFACE CAPABILITIES

Core Characteristics of RAN Function Exposure

RAN Function Exposure is the mechanism by which the E2 interface abstracts vendor-specific radio network functions into standardized, discoverable services. This enables xApps to monitor and control the RAN without knowledge of proprietary hardware implementations.

01

E2 Service Model (E2SM)

The E2 Service Model defines the standardized contract between an xApp and a specific RAN function. Each E2SM specifies the RAN Function Definition, supported Event Triggers, Actions, and Insert Indications.

  • E2SM-KPM: Key Performance Measurement collection and reporting
  • E2SM-RC: RAN Control for UE and cell-level configuration
  • E2SM-NI: Network Interface management and flow control
  • E2SM-CCC: Coordination between multiple xApps for conflict resolution
4+
Standardized E2SM Categories
02

RAN Function Definition

Each exposed RAN function publishes a RAN Function Definition that describes its capabilities in a machine-readable format. This definition includes the OID (Object Identifier) of the E2SM, supported event trigger styles, report styles, and insert styles.

  • Enables plug-and-play interoperability between any vendor's O-CU/O-DU and any xApp
  • The Near-RT RIC uses this definition to match xApp subscriptions with available RAN functions
  • Eliminates the need for proprietary adapter layers per vendor
03

Event Trigger Mechanisms

RAN Function Exposure supports multiple event trigger styles that determine when the RAN sends data to subscribing xApps:

  • Periodic Reporting: Configurable intervals (e.g., every 100ms) for continuous monitoring
  • On-Demand Request: xApp explicitly requests a single measurement report
  • Event-Triggered: RAN pushes data only when a defined threshold is crossed (e.g., PRB utilization > 80%)
  • Insert Indication: RAN suspends a procedure and awaits policy guidance from the xApp before proceeding
04

Subscription-Based Data Access

xApps interact with RAN functions through a publish-subscribe model over the E2 interface. An xApp sends an RIC Subscription Request specifying the E2SM, event trigger definition, and action definition.

  • The RAN node acknowledges with an RIC Subscription Response or rejects with a failure cause
  • Multiple xApps can simultaneously subscribe to the same RAN function
  • The RIC Indication message carries the actual measurement or event data from RAN to xApp
  • Subscriptions persist until explicitly deleted or the E2 connection is reset
05

Control Actions and Policies

Beyond monitoring, RAN Function Exposure enables near-real-time control through standardized action definitions. xApps issue RIC Control Requests that map to specific RAN procedures:

  • UE-Level Control: Modify per-UE radio bearer configuration, handover thresholds, or scheduling weights
  • Cell-Level Control: Adjust cell individual offsets, transmission power, or antenna tilt
  • Slice-Level Control: Reallocate resource block quotas between network slices
  • The RAN executes the action and returns a RIC Control Acknowledge with success or failure status
06

Insert Indication for Policy Guidance

The Insert Indication mechanism allows the RAN to pause a procedure and query an xApp for a decision before continuing. This enables policy-driven control loops without replacing the RAN's native procedures.

  • Example: Before admitting a new UE, the O-DU sends an Insert Indication to a QoE Optimization xApp
  • The xApp evaluates current load and UE priority, then returns an Insert Indication Response with admission or rejection
  • The RAN resumes its procedure using the xApp's guidance
  • This preserves the RAN's protocol stack integrity while externalizing decision logic
RAN FUNCTION EXPOSURE

Frequently Asked Questions

Essential questions about how the E2 interface abstracts and exposes RAN control and data collection services to xApps for vendor-agnostic optimization.

RAN Function Exposure is the capability of the E2 interface to abstract and expose specific RAN control and data collection services to xApps via a standardized, vendor-agnostic API. It defines what a RAN node can do and how an xApp can request those services, decoupling the optimization logic from the underlying hardware implementation. The mechanism relies on E2 Service Models (E2SMs), which are formal descriptions of exposed RAN functions—such as scheduling, beamforming, or KPI reporting—that both the RAN node and the Near-RT RIC understand. This exposure layer transforms proprietary base station capabilities into a set of discoverable, callable procedures, enabling a marketplace of third-party xApps to optimize any compliant RAN without custom integration.

Prasad Kumkar

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.