RF ML Model Training

An integrated software and hardware solution for training RF ML models. Includes DeepSig’s OmniSIG® Studio and Deepwave’s RF AI edge platform.

There are many barriers to training RF AI models. Not only are AI skills required, but working with RF data also requires specialized knowledge with a steep learning curve.

What’s more, there are few native, purpose-built RF AI tools available. Due to the immaturity of the RF AI ecosystem, the broader RF Machine Learning Operations (MLOps) tool chain consists of inoperable point solutions, making enterprise-grade RF AI model development accessible only to the most advanced researchers and scientists.

This turnkey, purpose-built RF ML Model Training solution accelerates the development, training, and deployment of advanced signal detection models. It also supports real-time spectrum awareness use cases, including:

• Autonomous 5G operation and optimization
• Next-generation dynamic spectrum sharing
• Foundational enabler for sense-native 6G

• ORAN interfaces
• Secure operation for defense wireless systems

How it Works

DeepSig’s OmniSIG Studio is an RF MLOps software platform. The product is specifically designed for editing, labeling, and curating RF waveforms to train deep learning models for signal detection and classification. OmniSIG Studio features a no-code interface that enables users with only basic RF knowledge to quickly develop, train, and deploy neural network models for complex RF waveform recognition tasks.

Deepwave provides an AI-enabled platform for turning RF data into RF intelligence. The platform is purpose-built for real-time RF deep learning and signal processing applications, and leverages tightly integrated software-defined radio (SDR) technology with high-performance computing. Deepwave’s main product lines are AIR-T edge hardware and AirStack software.

Solution Highlights

• Purpose-built RF AI system
• Ability to gather raw data and develop synthetic training data to refine models within an integrated system
• Human-in-the-loop analysis

• Rapid RF AI model development enabled by seamless RF MLOps workflow
• RF native auto data labeling
• Easy-to-use web UI
• Plug-and-go solution deployment