Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
|
| 2 |
`power_dbm` stored in the file attrs is the AWG-programmed power; apply the relevant net chain gain (see Sweep Grid above, or use `--attenuator_db`/`--gain_db` in the loading utility) to recover the actual power delivered to the MZM.
|
| 3 |
|
|
@@ -19,6 +93,7 @@ Raw captures were aligned, windowed, and processed into the released (`noisy_mod
|
|
| 19 |
A lightweight, open-source Python utility (`visualize_SS_Sine_data_shards.py`, MIT license) for loading and visualizing any data point by (frequency, power, repetition) is included in this repository.
|
| 20 |
|
| 21 |
### Requirements
|
|
|
|
| 22 |
```bash
|
| 23 |
pip install h5py numpy matplotlib
|
| 24 |
```
|
|
@@ -26,17 +101,20 @@ pip install h5py numpy matplotlib
|
|
| 26 |
### Usage
|
| 27 |
|
| 28 |
List all available (frequency, power) sweep points in a shard directory:
|
|
|
|
| 29 |
```bash
|
| 30 |
python visualize_SS_Sine_data_shards.py --shards_dir /path/to/shards --list
|
| 31 |
```
|
| 32 |
|
| 33 |
Load and plot a specific data point:
|
|
|
|
| 34 |
```bash
|
| 35 |
python visualize_SS_Sine_data_shards.py --shards_dir /path/to/shards \
|
| 36 |
--frequency 3.5 --power -10.0 --repetition 0 --window 0
|
| 37 |
```
|
| 38 |
|
| 39 |
To display the actual RF power delivered to the modulator (rather than the AWG-programmed value), pass the attenuator configuration used for that capture:
|
|
|
|
| 40 |
```bash
|
| 41 |
python visualize_SS_Sine_data_shards.py --shards_dir /path/to/shards \
|
| 42 |
--frequency 3.5 --power -10.0 --attenuator_db 30
|
|
@@ -45,6 +123,7 @@ python visualize_SS_Sine_data_shards.py --shards_dir /path/to/shards \
|
|
| 45 |
This plots three stacked subplots: the distorted capture (`noisy_mod`), the clean reconstructed target (`clean_mod`), and the shared per-repetition calibration block (`calib_unmod`).
|
| 46 |
|
| 47 |
The utility can also be used as a library:
|
|
|
|
| 48 |
```python
|
| 49 |
from visualize_SS_Sine_data_shards import load_point, plot_point
|
| 50 |
point = load_point("/path/to/shards", frequency=3.5, power=-10.0, repetition=0)
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-4.0
|
| 3 |
+
task_categories:
|
| 4 |
+
- time-series-forecasting
|
| 5 |
+
- other
|
| 6 |
+
tags:
|
| 7 |
+
- photonic-time-stretch
|
| 8 |
+
- analog-to-digital-converter
|
| 9 |
+
- rf-signal-processing
|
| 10 |
+
- spectrum-sensing
|
| 11 |
+
- signal-equalization
|
| 12 |
+
- denoising
|
| 13 |
+
- optics
|
| 14 |
+
- photonics
|
| 15 |
+
pretty_name: Photonic Time-Stretch ADC Equalization and Denoising Dataset
|
| 16 |
+
size_categories:
|
| 17 |
+
- 100K<n<1M
|
| 18 |
+
---
|
| 19 |
+
|
| 20 |
+
# Photonic Time-Stretch ADC Dataset for Equalization and Denoising
|
| 21 |
+
|
| 22 |
+
## Dataset Summary
|
| 23 |
+
|
| 24 |
+
This dataset provides 126,000 preprocessed captures (~1.4 TB) from a photonic time-stretch (PTS) analog-to-digital converter (ADC) testbed, collected across a systematic sweep of 56 RF frequencies and 15 power levels, with 150 repetitions per sweep point. Each data point contains a paired distorted measurement and undistorted clean target, along with an unmodulated baseline (calibration) window, enabling supervised training and benchmarking of machine learning models for PTS-ADC signal equalization, denoising, and spectrum sensing.
|
| 25 |
+
|
| 26 |
+
PTS-ADC front ends extend the usable digitization bandwidth of electronic ADCs well beyond their native limits by stretching RF signals in time using a chirped mode-locked laser pulse and a dispersive optical medium, before photodetection and electronic digitization. This architecture introduces its own nonlinear distortions (modulator nonlinearity, Gaussian pulse envelope effects, photodetector saturation), which this dataset is designed to help correct via machine learning.
|
| 27 |
+
|
| 28 |
+
## Supported Tasks
|
| 29 |
+
|
| 30 |
+
- **Signal equalization**: train models to recover the clean target waveform from the distorted measurement.
|
| 31 |
+
- **Denoising**: use the unmodulated calibration windows as paired baseline-noise segments, independent of equalization.
|
| 32 |
+
- **Spectrum sensing**: PTS-ADC front ends are central to wideband spectrum sensing in radar and wireless systems; this dataset supports training and evaluating ML-based spectrum sensing methods directly, across a swept frequency-power grid.
|
| 33 |
+
|
| 34 |
+
## Dataset Structure
|
| 35 |
+
|
| 36 |
+
### Sweep Grid
|
| 37 |
+
|
| 38 |
+
| Parameter | Value |
|
| 39 |
+
|---|---|
|
| 40 |
+
| RF frequency | 0.5 – 6 GHz (56 values) |
|
| 41 |
+
| Power level (at MZM) | -14.7737 to 13.2262 dBm (15 values) |
|
| 42 |
+
| Repetitions per sweep point | 150 |
|
| 43 |
+
| Total data points | 126,000 (56 × 15 × 150) |
|
| 44 |
+
|
| 45 |
+
The power level reported is the RF power delivered to the modulator (MZM), computed by applying the RF chain gain (RF amplifier, cable losses, and a fixed attenuator) to the power programmed at the AWG. Two attenuator configurations were used: a 30 dB attenuator (net chain gain 15.2263 dB) and a 20 dB attenuator (net chain gain 25.2263 dB). Files collected with the 20 dB attenuator are marked `high_power` in the filename.
|
| 46 |
+
|
| 47 |
+
### Data Fields
|
| 48 |
+
|
| 49 |
+
Each repetition is stored as an HDF5 group containing:
|
| 50 |
+
|
| 51 |
+
| Field | Description |
|
| 52 |
+
|---|---|
|
| 53 |
+
| `noisy_mod` | Distorted RF capture after the PTS path (the signal to be corrected) |
|
| 54 |
+
| `clean_mod` | Reconstructed undistorted ground-truth target, sample-aligned to `noisy_mod` |
|
| 55 |
+
| `calib_unmod` | Unmodulated (gap) calibration window, shared per repetition |
|
| 56 |
+
| `rf_phase` | Reference phase of the window, extracted via the Hilbert transform of the clean target |
|
| 57 |
+
|
| 58 |
+
### File Organization
|
| 59 |
+
|
| 60 |
+
Data is sharded into one HDF5 file per (frequency, power) sweep point, with repetitions stored as internal groups within each file. This keeps individual files at a manageable size and allows any single (frequency, power, repetition) data point to be accessed directly without downloading the full dataset.
|
| 61 |
+
|
| 62 |
+
```
|
| 63 |
+
shards_dir/
|
| 64 |
+
{stem}_point{NNNN}_f{frequency:.4g}GHz_p{power:+.1f}dBm[_high_power].h5
|
| 65 |
+
attrs: power_dbm, frequency_ghz, source_point_path,
|
| 66 |
+
pulses_per_window, pulse_samples
|
| 67 |
+
/rep_000/
|
| 68 |
+
├── calib_unmod 1D float32 (calib_len,)
|
| 69 |
+
├── clean_mod 2D float32 (n_windows, win_len)
|
| 70 |
+
├── noisy_mod 2D float32 (n_windows, win_len)
|
| 71 |
+
└── rf_phase 1D float32 (n_windows,)
|
| 72 |
+
/rep_001/...
|
| 73 |
+
...
|
| 74 |
+
```
|
| 75 |
|
| 76 |
`power_dbm` stored in the file attrs is the AWG-programmed power; apply the relevant net chain gain (see Sweep Grid above, or use `--attenuator_db`/`--gain_db` in the loading utility) to recover the actual power delivered to the MZM.
|
| 77 |
|
|
|
|
| 93 |
A lightweight, open-source Python utility (`visualize_SS_Sine_data_shards.py`, MIT license) for loading and visualizing any data point by (frequency, power, repetition) is included in this repository.
|
| 94 |
|
| 95 |
### Requirements
|
| 96 |
+
|
| 97 |
```bash
|
| 98 |
pip install h5py numpy matplotlib
|
| 99 |
```
|
|
|
|
| 101 |
### Usage
|
| 102 |
|
| 103 |
List all available (frequency, power) sweep points in a shard directory:
|
| 104 |
+
|
| 105 |
```bash
|
| 106 |
python visualize_SS_Sine_data_shards.py --shards_dir /path/to/shards --list
|
| 107 |
```
|
| 108 |
|
| 109 |
Load and plot a specific data point:
|
| 110 |
+
|
| 111 |
```bash
|
| 112 |
python visualize_SS_Sine_data_shards.py --shards_dir /path/to/shards \
|
| 113 |
--frequency 3.5 --power -10.0 --repetition 0 --window 0
|
| 114 |
```
|
| 115 |
|
| 116 |
To display the actual RF power delivered to the modulator (rather than the AWG-programmed value), pass the attenuator configuration used for that capture:
|
| 117 |
+
|
| 118 |
```bash
|
| 119 |
python visualize_SS_Sine_data_shards.py --shards_dir /path/to/shards \
|
| 120 |
--frequency 3.5 --power -10.0 --attenuator_db 30
|
|
|
|
| 123 |
This plots three stacked subplots: the distorted capture (`noisy_mod`), the clean reconstructed target (`clean_mod`), and the shared per-repetition calibration block (`calib_unmod`).
|
| 124 |
|
| 125 |
The utility can also be used as a library:
|
| 126 |
+
|
| 127 |
```python
|
| 128 |
from visualize_SS_Sine_data_shards import load_point, plot_point
|
| 129 |
point = load_point("/path/to/shards", frequency=3.5, power=-10.0, repetition=0)
|