With RFix
- Place emitters and a receiver on the RF map.
- Connect signals, interference, and ITU-R or Sionna paths.
- Inspect the IQ and export labels, SigMF, and the source project.
RFix vs TorchSig
A complete visual RF workflow vs a code-first RFML toolkit.
RFix takes teams from drag-and-drop RF design and terrain-aware propagation to analyzed, labeled IQ without assembling Python layers. TorchSig provides useful PyTorch signal generators, impairments, labels, and datasets, but its core workflow remains code-first and needs separate spatial, analysis, and agent tooling.
RFix advantage: Faster visual, terrain-aware scenarios and traceable IQ datasets in one product.
Choose TorchSig mainly for: Code-first PyTorch RFML datasets, transforms, and training.
Why choose RFix
The easier end-to-end RF workflow.
RFix removes the integration work between scenario design, propagation, IQ analysis, datasets, and automation.
You want drag-and-drop signal construction, map placement, propagation, analysis, and export without building those layers in Python.
You need ITU-R or Sionna terrain workflows and multi-emitter interference tied to labeled IQ and SigMF.
You want the same project usable by engineers, batch APIs, and MCP-connected AI agents.
Workflow advantage
What RFix includes in one product.
| Capability | RFix | TorchSig |
|---|---|---|
| Product scope | End-to-end workspace for building, locating, analyzing, and exporting RF scenarios. | RFML library for generating and transforming PyTorch training data.[1] |
| Signal library | 94+ fully included visual signal nodes in one packaged workspace. | 57 built-in low-level signal types across 10 modulation families.[2] |
| Visual authoring | Drag-and-drop graph with reusable projects; routine scenarios need no Python. | Dataset construction uses Python, transforms, data loaders, and DatasetCreator; a beta GUI is available.[3][6] |
| Maps & terrain | Place emitters, receivers, motion, and terrain on a map tied to the signal graph. | No built-in RF map, terrain, or emitter/receiver placement workflow.[1] |
| Propagation & interference | ITU-R models, Sionna terrain paths, and multi-emitter interference. | Strong impairments and channel transforms, but no ITU-R, Sionna, or spatial interference workflow.[5] |
| IQ import & analysis | Open raw IQ, WAV-style data, or SigMF and inspect spectrogram and time-domain views. | Outputs IQ or spectrogram data; inspection and custom analysis use Python or notebooks.[4] |
| Labels & deliverables | Traceable annotations, raw IQ, SigMF metadata, and portable visual projects. | Flexible labels and iterable or HDF5-backed PyTorch datasets.[4] |
| RF hardware | Built-in SDR I/O plus paired VSG/analyzer control, waveform playback, measurements, IQ capture, and SCPI. | No built-in SDR, VSG, signal-analyzer, or SCPI control workflow.[1] |
| Automation & AI agents | Batch APIs and native MCP tools plan, generate, inspect, and export RF projects. | No native MCP or RF-scenario agent workflow; automation uses the Python APIs.[3] |
| Setup & operation | Packaged desktop and web interfaces for design, generation, analysis, and export. | Python/PyTorch, notebooks, or Docker; advanced workflows are assembled in code.[3] |
| Open-source fit | Commercial platform with free and licensed capabilities. | MIT-licensed for teams that want to modify or embed the RFML library.[3] |
Concrete workflow
How RFix simplifies: Build a terrain-aware multi-emitter interference dataset.
With TorchSig
- Configure generators, transforms, and labels in Python.
- Add spatial and standards-based propagation separately.
- Analyze and package the resulting dataset with Python tooling.
Specialist exception
When TorchSig is the specialist choice.
Choose TorchSig primarily when one of these specialist requirements matters more than an integrated scenario-to-IQ workflow:
- You want an open-source, code-first dataset toolkit that lives inside PyTorch workflows.
- You need reusable signal builders, impairments, metadata transforms, or HDF5-backed RFML datasets.
- You are comfortable building interactive analysis, spatial modeling, and agent integration around the library.
The short version
RFix removes the integration work between visual design, maps, propagation, interference, IQ analysis, traceable exports, APIs, and MCP. Choose TorchSig mainly when the deliverable is a code-first PyTorch RFML pipeline rather than an end-to-end RF project.
See it on your own scenario
The fastest way to compare is on your data. We can generate a sample labeled dataset from a scenario like yours so you can judge the fit directly.
Explore RFix
See the workflow in context.
Verification
Official sources.
Published ยท Last reviewed