From your 96 × 14 × 10000 dataset, are currently extracting:
| Category | Number of Parameters |
|---|---|
| Channel Impulse Response | 96 (delay taps) |
| Complex Gains | 96 (per path) |
| Multipath Delays | 31 paths + 3 metrics |
| Doppler Parameters | 5 metrics |
| Angle of Arrival | 23 angles + 2 metrics |
| Coherence Parameters | 12 metrics |
| Statistics | 8 metrics |
| TOTAL per sample | ~270 parameters |
- Total stored parameters: ~2.7 million (270 × 10,000)
- Visualization outputs: 12 plots per sample
With data structure: 10,000 samples × 96 subcarriers × 14 symbols, we could extract:
- Impulse response taps: 96
- Complex gains: 96 (magnitude + phase = 192 parameters)
- Power delay profile: 96
- RMS delay spread: 1
- Mean excess delay: 1
- Maximum excess delay: 1
- Delay window: 1
- Number of significant taps: 1
- Tap locations: up to 96
Subtotal: ~400 parameters/sample
- Frequency response: 96 (magnitude + phase = 192)
- Coherence bandwidth (multiple thresholds): 5
- Frequency correlation matrix: 96×96 = 9,216
- Subcarrier correlation coefficients: 96
- Group delay: 96
- Phase linearity: 1
Subtotal: ~9,600 parameters/sample
- Doppler spectrum: 14 points
- Maximum Doppler shift: 1
- Doppler spread: 1
- Coherence time: 1
- Channel aging rate: 1
- Time correlation matrix: 14×14 = 196
- Symbol-to-symbol correlation: 14
- Channel variation rate: 1
- Doppler frequency components: up to 14
Subtotal: ~230 parameters/sample
- Distribution parameters (Rayleigh/Rician): 3
- K-factor: 1
- Nakagami-m parameter: 1
- Fading severity: 1
- Level crossing rate: 1
- Average fade duration: 1
- Higher-order statistics: 5
- Channel capacity: 1
- Outage probability: 1
Subtotal: ~15 parameters (global)
- AoA spectrum: 181 angles
- AoA peaks: up to 96
- Angular spread: 1
- Spatial correlation: 96×96 = 9,216
- Virtual array manifold: 96×181 = 17,376
- MUSIC pseudospectrum: 181
- Beamforming weights: 96×181 = 17,376
Subtotal: ~44,000 parameters/sample
- Instantaneous capacity: 1
- Ergodic capacity: 1
- Outage capacity: 1
- Mutual information: 1
- Channel entropy: 1
- Diversity order: 1
- Multiplexing gain: 1
Subtotal: ~7 parameters/sample
- Time-frequency images: 96×14 = 1,344 pixels
- STFT representations: 96×14×2 = 2,688
- Wavelet coefficients: up to 96×14×4 = 5,376
- PCA components: up to 96
- Autoencoder features: configurable
- Channel state information (CSI) matrices: 96×14×2 = 2,688
Subtotal: ~12,000 parameters/sample
| Level | Parameters per Sample | Total for 10k Samples |
|---|---|---|
| Basic (current) | 270 | 2.7M |
| Comprehensive | ~66,000 | 660M |
| Maximum (with all features) | ~150,000 | 1.5B |
| Metric | Value |
|---|---|
| Currently extracted | ~270 parameters/sample |
| Should extract | ~1,400 parameters/sample |
| Maximum possible | ~66,000 parameters/sample |
| Your data capacity | 96×14×10000 = 13.44M complex values |
| Theoretical information | 26.88M real values |