RF Data Notes
RF Data NotesBy Paul SorensenJanuary 20, 20263 min read

RF data is not ordinary file data

Large IQ captures carry value only when the context stays with them. Metadata, preview, access history, and analysis links matter as much as the file itself.

An IQ capture is a file plus context

RF data does not behave like ordinary file data. A document can often stand on its own. A photo usually carries enough context for a person to understand what it is. A log file might be ugly, but the structure is usually readable.

An IQ capture is different. The file may be huge, and the contents may not be obvious without the right tools. The useful context often lives somewhere else: center frequency, sample rate, bandwidth, timestamp, sensor configuration, collection environment, operator notes, processing steps, and whatever happened after the capture.

If that context gets separated from the file, the data becomes harder to use.

From IQ capture to reusable RF data assetA raw IQ capture becomes a reusable RF data asset when context such as metadata, preview, access history, and analysis outputs stays attached to it.From IQ capture to reusable RF data assetThe useful asset is the capture plus the context around it.Raw IQ capturelarge sample fileContext that stays attachedcenter frequencysample ratebandwidthtimestampsensor and configspectrogram previewaccess historyanalysis outputsSearchable RFdata assetfind, preview, reuse
An IQ capture becomes more useful when metadata, preview, access history, and analysis outputs stay attached to it.

The pain shows up later

The problem usually does not show up on day one. It shows up later, when someone asks for the old capture from a test, wants to compare a new signal against prior data, or needs to reproduce an analysis from months ago.

Then the questions start. Which file is the right one? What configuration produced it? Was this the raw capture or a processed version? Can we preview it before downloading the whole thing? Has anyone already analyzed it? Who had access to it? Can another team reuse it?

Many RF teams answer those questions with a mix of folders, naming conventions, spreadsheets, notebooks, object storage, scripts, and memory. That can work for a while.

Duplication is a workflow symptom

Someone finally finds the capture they need on a shared drive, so they copy it into their project folder to keep it for later. Another engineer does the same thing a week later. A third person saves a local copy before running analysis because they do not want to lose track of it again.

Nobody is doing anything wrong. They are protecting their own workflow.

But with IQ sample data, those just-in-case copies add up fast. The same large capture can end up duplicated across project folders, personal scratch space, shared drives, and analysis directories. Storage fills up, nobody knows which copy is canonical, and the team still has the original search problem.

The keep it for later problemEngineers copy large IQ captures into multiple folders because they do not want to lose track of useful files.The "keep it for later" problemCopying a file can be rational for one person and costly for the team.Shared driveoriginal IQ captureproject folderanalyst machinescratch spaceanalysis directoryResultstorage growscanonical file gets unclearsearch problem remains
When teams cannot reliably find and trust prior captures, people copy them "just in case." With large IQ captures, that workaround gets expensive fast.

What needs to stay attached

RF data needs a stronger data layer because large signal datasets become more useful when the metadata, preview, access control, compression, and analysis trail stay connected to the capture.

A better RF data workflow should make it easier to ingest large captures, attach and search useful metadata, preview signals without moving the full file, reduce storage burden where possible, control who can access sensitive data, preserve the link between raw captures and derived analysis, and deploy in environments that cannot rely on public cloud services.

That is the direction we are taking with SigDrive: an RF data lake for teams that need large captures to remain findable, understandable, and reusable after the original collection is over.

Working through this problem?

We are opening early conversations with RF teams dealing with large captures, duplicated files, messy metadata, preview pain, and secure deployment constraints.