Studio Signal Generator¶
Open Signal Generator in the sidebar, or choose Get Started → Signal Generator. The first visit generates a private 5G NR numerology preview. Choose one of the five signal families, select a preset, and press Generate & preview. Parameter edits mark the current plots stale and disable waveform export and dataset creation until generation succeeds.
The onboarding dialog has exactly one highlighted action: Signal Generator. Existing datasets are the second choice, and CSV upload is third. PA and DPD each have one model workspace with Training and Testing tabs. Existing task URLs remain valid. Testing displays the selected dataset version's test split count and equivalent time, before any model warm-up or evaluation-edge exclusions.
Waveforms and coverage¶
The generator provides synthetic engineering stimuli with standard numerologies, not complete protocol implementations or certified reference test models.
| Family | Presets | Implemented signal |
|---|---|---|
| 5G NR | FR1 20 / 100 MHz; FR2 100 MHz | 30 / 120 kHz subcarrier spacing, normal CP, 51 / 273 / 66 resource-block-sized payload grids, generic pilots |
| Wi-Fi 6 | 20 / 40 / 80 / 160 MHz | 78.125 kHz spacing, 0.8 µs guard interval, up to 1024-QAM |
| Wi-Fi 7 | 20 / 40 / 80 / 160 / 320 MHz | 78.125 kHz spacing, up to 4096-QAM |
| Wi-Fi 8 | 80 / 160 / 320 MHz | Experimental OFDM numerology profile; generic allocations and pilots |
| Custom | OFDM/OFDMA, single-carrier QAM/PSK, single tone, multitone, chirp | Fully editable baseband parameters |
OFDM payload symbols are uncoded and continuous. Pilots are generic seeded BPSK, including when their bin positions are explicitly specified. Synchronization, FEC, NR physical/control-channel mapping, WLAN preambles, standard RU allocation bitmaps, MAC packets, and draft-specific UHR mechanisms are not implemented. The GUI, saved metadata, and exports all disclose this scope. Changing preset numerology marks the waveform custom. Existing known-waveform evaluation bindings are not assigned to these generated signals.
NR CP timing follows TS 38.211, §5.3.1. The long normal CP appears twice per subframe; extended CP requires 60 kHz spacing. For WLAN background, see the IEEE 802.11 working group. As of 2026-09-13, 802.11bn remains a draft; its development status is tracked by IEEE TGbn.
Advanced parameters¶
- Output sample rate, nominal baseband bandwidth, RF carrier metadata, RMS and seed.
- Exact complex sample count, or equivalent duration rounded to the nearest sample. One complex sample contains I and Q. The current limit is 256–1,000,000 samples.
- FFT size, 1/2/4/8× oversampling, subcarrier spacing, fixed/NR CP and DC null.
- Up to 16 OFDMA channels with independent carrier counts, modulation and relative power. Counts include pilots. Channel gaps are specified in FFT bins. These are users within one RF band, not separate RF carriers for adjacent-channel metrics.
- Per-channel pilot comb, explicit signed carrier indices, or no pilots; pilot boost.
- RRC QAM/PSK pulse shaping, samples per symbol, roll-off and filter half-span.
- Tone frequency, multitone count, and a linear chirp over the declared bandwidth.
- Frequency offset, I gain mismatch, Q phase mismatch, I/Q DC offsets, envelope clipping and independent white Gaussian noise.
SCS = output sample rate / (FFT size × oversampling). The FFT and fixed CP inputs
refer to the grid before oversampling; measured CP lengths are reported in output
samples. Changing SCS in the GUI updates the output sample rate. Carrier frequency
is saved as RF metadata and never digitally mixes a GHz signal into the baseband.
RMS normalization precedes impairments. I gain and Q phase mismatch are applied first, followed by DC offset, frequency offset, clipping and noise. There is no post-impairment normalization or hidden receiver equalization.
Visualizations and measurements¶
Time plots display the first 2,048 contiguous samples without strided sampling. Spectrum, RMS, peak, PAPR and CCDF statistics use the entire exported waveform. Welch PSD uses a Hann window, 50% overlap and density scaling. Power is referenced to unit complex RMS, not calibrated watts or dBm. Occupied bandwidth contains the central 99% of integrated Welch power, including impairments.
PAPR is the maximum sample power divided by average sample power, in dB. It does not estimate unsampled analog peaks. CCDF is the empirical exceedance probability of power above its mean; zero-probability points are omitted on the logarithmic plot, with empirical resolution 1/N.
OFDM constellation and diagnostic reference EVM use data carriers from up to the first 16 complete symbols. The FFT receiver uses the known generation normalization; it does not fit gain, phase, delay, or equalization. A capture with no complete symbol has no reference EVM. Single-carrier constellation shows transmitted symbols before RRC filtering and impairments, not a receiver estimate. Incomplete final symbols are retained to preserve the exact requested sample count and disclosed in metadata.
Export and training¶
Save configuration downloads JSON; Load configuration validates it against
the server contract before applying it. Export I/Q + configuration contains
iq.csv (I,Q), iq.npy (float32 N×2), configuration, measurements, source hash,
NumPy version and scope notes. CSV float32 values round-trip exactly. Seeded
regeneration requires the recorded configuration, generator implementation and
numeric environment. Generated records live privately under signals/sg-<sha>/
inside the workspace.
The waveform alone has no measured PA output. Create synthetic PA dataset adds a tunable illustrative nonlinear memory PA response. It requires at least 8,192 samples and at least 256 samples per split. Users can configure gain, compression, AM/PM, memory, output noise, split fractions and guards. The exact generated input is preserved, and both input and output are marked SYNTHETIC. The resulting dataset opens the PA Model workspace and can use the existing DPD workflow after PA training.
Generated datasets are private by default. Optional publication uses the existing dataset contribution workflow: an explicit reviewed package, a separate branch, push and PR, followed by human merge review. Contact: emi.lab@outlook.com.
Hosted sessions apply authentication, tenant isolation, size limits and generation quotas. Dataset creation respects the host's custom-dataset capability. No RF transmitter, GitHub submission, or email is activated by generating a waveform.
Verification¶
Numerical tests cover all 20 presets, exact sample counts, one-millisecond NR CP timing, FFT allocations, empty bins, deterministic impairments, analytic single-tone PAPR, invalid parameters, export round trips and dataset provenance. API tests cover authentication, CSRF, host feature gating and per-version test counts. Hosted tests verify that one session cannot access another session's generated signals or data.
scripts/verify_signal_generator.mjs exercises the real local browser at 1366×768
and 1920×1080: onboarding order and highlight, generation, stale-result protection,
custom two-channel OFDMA with explicit pilots and noise, ZIP export, dataset creation,
and real CPU PA/DPD training and testing. These checks validate the software workflow;
they do not constitute independent standards conformance or physical RF validation.