Testes de IA

The AI-RAN 

Prove AI-RAN Works Under Real-World Conditions

Whether AI runs on the gNodeB, on the device, or through the cloud-native RAN Intelligent Controller (RIC) and its xApps, rApps, and aApps, it must be proven before it reaches a live network. VIAVI validates it from the radio environment it runs in, through the decisions it makes, to the workload it carries and the service that proves the result.

We are a member of the AI-RAN Alliance and chair its Data-for-AI (Data4AI) Task Group, helping define how AI-RAN performance is measured across the industry.

Prove AI-RAN Works Under Real-World Conditions

A true-to-life test environment

AI-RAN behavior depends on real radio conditions: coverage, beams, reflections, and interference. Validation runs against city-scale Digital Twins built with ray-tracing technology and calibrated with real field measurements, testing the AI-driven features operating inside the radio itself, including beam management, mobility optimization, interference handling, and ISAC. TeraVM AI RSG generates the 3D scenarios and training datasets behind those tests, accelerated on NVIDIA GPUs.

RAN intelligence making the calls

Automated decisions like energy efficiency, MU-MIMO interference management, QoS optimization, traffic steering, and handover parameter optimization run through xApps and rApps on the RIC, and they need validating too. TeraVM AI RSG scores those decisions automatically.

Real devices and traffic, at scale

AI-RAN must hold up under real UE behavior, mobility, and signaling, across many cells. TM500 with ray tracing emulates all three at city scale, and the Vertex Channel Emulator reproduces the same ray-traced channel for the base station under test.

The AI workload itself

AI responses must be correct, consistent, and timely under live RAN conditions, on the same shared GPU and CPU infrastructure the network runs on: LLM inference, XR, video analytics, and other intelligent applications operating alongside live RAN functions. AI Inference Test measures the model directly at the edge or core, with CyberFlood driving the workload around it for full end-to-end performance.

Testing, run as a service

Some operators need deployment-grade results delivered independently, with enough diagnostic depth to act fast when something fails. That's VALOR: VIAVI's lab as a service, running AI-Driven Lab Intelligence across diagnostics, real-time health monitoring, predictive analytics, topology mapping, smart test orchestration, and an AI Experts interface for lab queries.

The new AI-RAN attack surface

AI-RAN introduces a new attack surface: models, APIs, and shared infrastructure. TeraVM Security tests it directly.

Speed, powered by NVIDIA acceleration

Generating scenarios, running tests, and returning results all take compute. NVIDIA-accelerated GPUs cut the time each step takes, so validation keeps pace with AI-RAN.

Automate the Engineer Workflow

AI Experts with Agents to increase engineer productivity

AI Experts are purpose-built productivity capabilities running on the NITRO Wireless platform. They automate configuration, analysis, and diagnostics across field and lab workflows.

In the field

AI Experts deliver contextual guidance from wireless standards, instrument functionality, and real-world signal behavior, directly on the instrument, so engineers validate faster and with greater accuracy.

In the lab

AI Experts cut the path from provisioning to validated results. Test setup, diagnostic triage, and topology awareness are handled automatically, so engineering teams hit tighter timelines without adding headcount.

AI Experts run today on OneAdvisor 800 Wireless, TM500, and TeraVM

• Generate log masks on demand. What previously took days resolves in seconds.
• Provide 24/7, multi-lingual guidance so engineers find answers immediately, without waiting on support queues or searching documentation.
• Cut the time spent tracing issues to root cause. Faster diagnosis means faster resolution and less stalled test time.
• Generate automated test cases, reducing the manual effort of building and maintaining test coverage from scratch.
• Run security threat testing automatically, validating network resilience without manual test construction.

OneAdvisor 800 Wireless AI Expert

Contextual guidance for field engineers: RF and signal expertise on demand, multilingual support, and real-world signal interpretation.

All of it runs on VAMOS – Automation and Management Orchestration System, the interface that hosts the AI Experts suite and enables AI-driven orchestration of automated test workflows across multi-vendor environments.

Redes mais inteligentes com IA

A VIAVI oferece ferramentas para validar, otimizar e dimensionar a IA em todo o ciclo de vida da RAN, partindo do AIRtoCore™.
Eliminamos as incertezas da implantação de IA, construindo maior confiança por meio de:

Smarter Networks with AI

Teste de IA na sua rede 

A próxima geração de redes de acesso via rádio exige mais: alocação dinâmica de recursos, entregas mais inteligentes e menor uso de energia. A IA torna isso possível.  

Quer seus algoritmos de IA sejam aplicados diretamente no gNodeB, dispositivo ou usados ou com o controlador inteligente da RAN (RIC) nativo da nuvem que usa aplicativos orientados por IA (xApps, rApps, aApps), a VIAVI conta com soluções de pesquisa, treinamento e teste para garantir que essa nova tecnologia funcione, interopere e melhore o desempenho de rede. Podemos até avaliar que as tecnologias RAN e IA podem funcionar com eficiência na mesma infraestrutura.  

Fornecimento de soluções de IA para produtividade: AI Test Experts  

O papel da IA não é substituir pessoas, mas tornar o trabalho mais rápido, inteligente e eficiente. Na VIAVI, estamos construindo a ferramenta AI Test Experts que reduz tarefas repetitivas, automatiza fluxos de trabalho e melhora a usabilidade.