TeraVM AI RSG AI RAN Scenario Generator and RAN Digital Twin Platform

Software-based RAN digital twin that simulates, develops and validates 4G, 5G, 6G and NTN network behavior and xApps/rApps without physical infrastructure.

Cells: up to 5,000 cells
Cells per reference server: between 1,000 and 5,000 cells per reference server
UEs: 10,000 UEs
Simulation speed: 100x real-time

Datasheet

Overview

The VIAVI TeraVM AI RSG (AI RAN Scenario Generator) is a software-based RAN digital twin platform. It simulates system-level RAN behavior across 4G, 5G, emerging 6G and non-terrestrial networks, both open and purpose-built, so network behavior and optimization applications can be developed, trained and validated without physical infrastructure.

The platform can emulate a range of topologies and morphologies, configurations, mobility patterns and traffic profiles on real or synthetic maps. Layer 1 radio conditions are modeled and fed into a Layer 2 scheduler, which assigns Physical Resource Blocks based on the Channel Quality Indicator. Traffic demand profiles let UEs switch between idle and active states while using varying amounts of PRBs.

Software ecosystem

AI RSG combines four layers: the AI RSG simulation engine, the Python-based ADK (Application Development Kit), the AVE (App Validation Engine) and an embedded Claude AI assistant for natural-language access. The ADK lets app developers emulate app environments within the AI RSG ecosystem and test applications in closed loop, without integrating with a real NMS or RIC platform. AVE provides KPI scoring, benchmarking and goal-fulfilment analysis, and uses a Graph Neural Network model to predict optimal network configurations as a reference for the app under test. It can also issue pass/fail verdicts for CI pipelines.

Data generation and validation

AI RSG produces RAN-specific datasets for training AI/ML models used in SON, xApps and rApps, and supports benchmarking of RAN Intelligent Controllers in near-real-time and non-real-time scenarios. KPIs for cell, UE, QoS and massive MIMO are provided according to 3GPP and O-RAN Alliance specifications as well as VIAVI proprietary KPIs, and are exported through the supported interfaces to the application or platform under test.

Deployment and integration

The software runs in Docker containers in cloud environments (GCP, Azure, AWS) or on dedicated servers, including air-gapped on-premise installations. Native E2AP, O1 NETCONF and A1 interfaces allow integration with existing SMO, RIC and Grafana infrastructure. The digital twin can be calibrated with live data from field instruments and cell traces.

Key Features

RAN digital twin simulating system-level RAN behavior for 4G, 5G, emerging 6G and NTN networks
Open RAN validation: benchmark RICs in near-real-time and non-real-time scenarios
Generation of RAN-specific datasets for training AI/ML models for SON, rApps and xApps
Modeling of diverse topologies, scenarios, UE types, mobility models, traffic profiles and GIS maps, with topology import
Import of real-world maps with streets, buildings and network configurations
3D GIS scenarios and 3GPP mMIMO / AODT raytracing
Configurable UE type, cell type, frequency and traffic profile
Anomaly emulation, e.g. throughput reduction in specific cells
Python ADK for RAN control and KPI querying, with seed-based scenario replay
Simulation at 100x real-time with thousands of parallel scenarios
App Validation Engine (AVE) with automated KPI scoring, benchmarking and CI pass/fail verdicts
AI-based AVE using Graph Neural Network modeling to predict optimal network configuration
Embedded Claude AI assistant for natural-language scenario configuration, ADK code generation and results analysis
Native E2AP, O1 NETCONF and A1 interfaces with integration into existing SMO, RIC and Grafana infrastructure
Real-world calibration using live data from field instruments and cell traces
Cloud-native deployment on GCP, Azure, AWS or air-gapped on-premise
Kubernetes and container-ready, lightweight container deployment

Specifications

Scale

Cells	up to 5,000 cells Datasheet; the web page states up to 100,000 cells
Cells per reference server	between 1,000 and 5,000 cells per reference server
UEs	10,000 UEs Datasheet also states up to tens of thousands of UEs; the web page states 1 million+ UEs
Simulation speed	100x real-time

Network Technologies

Supported networks	4G, 5G, emerging 6G and NTN networks
UE types	Pedestrian, car, bus, drone, indoor etc.
Cell types	indoor, urban, rural

Interfaces and Compliance

Interface testing	E2, A1, O1
O-RAN interfaces	E2AP, O1 NETCONF, and A1
Spec compliance	O-RAN WG3 Spec Compliance
KPI categories	Cell, UE, QoS and mMIMO According to 3GPP specs, O-RAN Alliance specs and VIAVI proprietary KPIs

Deployment

Deployment	Docker containers with cloud environments or on dedicated servers
Cloud compatibility	GCP, Azure, AWS
Containerization	Kubernetes and container-ready
On-premise	air-gapped on-premise for defence environments
Deployment time	under an hour

Specifications are taken from the manufacturer's documentation. See the datasheet for complete and current values.

Applications

xApp and rApp development and validation
RAN Intelligent Controller (RIC) benchmarking
O-RAN interface compliance testing
AI/ML model training dataset generation
SON application testing
Massive MIMO and beam management optimization
Energy efficiency optimization
Non-terrestrial network simulation
Continuous integration testing of RAN applications
Autonomous network (L0–L5) validation
RAN vendor and CSP development labs
Research and defense environments

Options & Accessories

Option	Description
TVM6235	AI RSG Single User License
TVM6202	AI RSG Data License (100 Nodes, 300 Cells, 2000 UEs, 200k Reports/s)
TVM6203	AI RSG Advanced RF Model