ruvnet/RuView · error · CSIProcessingError

Failed to preprocess CSI data

Error message

Failed to preprocess CSI data: {e}

What it means

CSIProcessingError raised by CSIProcessor.preprocess_csi_data when any exception escapes its internal steps (_remove_noise, _apply_windowing, _normalize_amplitude). It is a wrapper: the original exception is stringified into the message, so inspect the embedded text to find the real cause. The wrapper exists so all preprocessing failures share one catchable type.

Solutions

1. Read the embedded message: `str(e)` contains the original exception — fix that root cause first.
2. Validate CSIData shape before calling: non-empty 2-D arrays with matching amplitude/phase shapes (see validationCode).
3. Drop or sanitize NaN/inf: `np.nan_to_num` on amplitude/phase, or reject the packet at ingestion.
4. If data comes from a capture buffer, wait until at least window_size samples exist before preprocessing.

Example fix

# before
clean = processor.preprocess_csi_data(csi_data)  # CSIProcessingError: Failed to preprocess CSI data: ...empty...

# after
if csi_data.amplitude.ndim == 2 and csi_data.amplitude.shape[0] >= processor.window_size:
    clean = processor.preprocess_csi_data(csi_data)
else:
    logger.warning('skipping undersized CSI batch %s', csi_data.amplitude.shape)

def preprocessable(csi_data, min_samples: int) -> bool:
    amp = getattr(csi_data, 'amplitude', None)
    pha = getattr(csi_data, 'phase', None)
    return (
        amp is not None and pha is not None
        and getattr(amp, 'ndim', 0) == 2 and getattr(pha, 'ndim', 0) == 2
        and amp.shape == pha.shape
        and amp.shape[0] >= min_samples
        and bool(np.isfinite(amp).all() and np.isfinite(pha).all())
    )

if preprocessable(csi_data, processor.window_size):
    clean = processor.preprocess_csi_data(csi_data)

try:
    clean = processor.preprocess_csi_data(csi_data)
except CSIProcessingError as e:
    logger.warning('dropping bad CSI batch (%s): %s',
                   getattr(csi_data, 'timestamp', '?'), e)
    continue  # skip this batch, keep the stream alive
# Do not catch blindly: inspect str(e) root cause before dropping

When it happens

Trigger: Calling processor.preprocess_csi_data(csi_data) with malformed CSIData: empty arrays (size 0 amplitude/phase), arrays with different subcarrier counts between amplitude and phase, 1-D phase where 2-D is expected, or NaN/inf values that break normalization (division by zero norm).

Common situations: Feeding the first packet from a cold ESP32 capture where the buffer is empty; simulators generating mismatched amplitude/phase shapes; CSI data truncated by packet loss so last window is ragged; NaNs introduced upstream by log of zero or division by zero in the extractor; test fixtures with synthetic shapes that violate the 2-D (samples x subcarriers) contract.

Related errors

• Failed to extract features
• Failed to detect human presence
• Pipeline processing failed
• ESP32 CSI data contains non-numeric values
• ESP32 CSI data incomplete: expected

AI-assisted analysis of ruvnet/RuView@4685618388 (2026-08-16). Data as JSON: /api/errors/edbeaf5cd2afd779. Report an issue: GitHub.

Appendix: source

Thrown at archive/v1/src/core/csi_processor.py:147

            CSIProcessingError: If preprocessing fails
        """
        if not self.enable_preprocessing:
            return csi_data
        
        try:
            # Remove noise from the signal
            cleaned_data = self._remove_noise(csi_data)
            
            # Apply windowing function
            windowed_data = self._apply_windowing(cleaned_data)
            
            # Normalize amplitude values
            normalized_data = self._normalize_amplitude(windowed_data)
            
            return normalized_data
            
        except Exception as e:
            raise CSIProcessingError(f"Failed to preprocess CSI data: {e}")
    
    def extract_features(self, csi_data: CSIData) -> Optional[CSIFeatures]:
        """Extract features from CSI data.
        
        Args:
            csi_data: Preprocessed CSI data
            
        Returns:
            Extracted features or None if disabled
            
        Raises:
            CSIProcessingError: If feature extraction fails
        """
        if not self.enable_feature_extraction:
            return None
        
        try:
            # Extract amplitude-based features

View on GitHub (pinned to 4685618388)