Implement TDOA/FDOA processing for multi-static radar measurements

[jonnyspicer]

Overview

Implement Time Difference of Arrival (TDOA) and Frequency Difference of Arrival (FDOA) processing to enable proper multi-static radar positioning from raw delay/Doppler measurements.

Problem Statement

Current Gap: The system expects individual radar delay/Doppler measurements but multi-static localization requires relative measurements between radar pairs.

From Flowchart: The "Solver" component shows "node pair detection" processing, indicating the need for relative measurements between radar nodes.

Current vs Required Processing

Current Approach

Radar1: delay=100ms, doppler=50Hz
Radar2: delay=120ms, doppler=30Hz
Radar3: delay=80ms,  doppler=70Hz
         ↓
Direct geometric localization (may be suboptimal)

Required Multi-Static Approach

Raw measurements → TDOA/FDOA computation → Localization
TDOA_12 = delay1 - delay2 = -20ms
TDOA_13 = delay1 - delay3 = +20ms  
FDOA_12 = doppler1 - doppler2 = +20Hz
FDOA_13 = doppler1 - doppler3 = -20Hz
         ↓
Optimized position solving with relative measurements

Target State for MVP

Phase 1: Basic TDOA Processing

1. Compute relative delays between radar pairs
2. Apply clock synchronization corrections (if available)
3. Interface with existing localization algorithms
4. Handle missing/invalid measurements gracefully

Phase 2: FDOA Integration (if needed)

1. Compute relative Doppler shifts for velocity estimation
2. Integrate with tracking filter for improved state estimation

Implementation Approach

class TdoaFdoaProcessor:
    def process_measurements(self, radar_detections, target_associations):
        """Convert individual radar measurements to relative TDOA/FDOA"""
        processed_targets = {}
        
        for target_id, detections in target_associations.items():
            if len(detections) >= 2:
                # Compute all pairwise TDOA/FDOA measurements
                tdoa_measurements = self._compute_tdoa_pairs(detections)
                fdoa_measurements = self._compute_fdoa_pairs(detections)
                
                processed_targets[target_id] = {
                    'tdoa': tdoa_measurements,
                    'fdoa': fdoa_measurements,
                    'radar_positions': [det['radar_pos'] for det in detections]
                }
        
        return processed_targets

Files to Create

• event/algorithm/processing/TdoaFdoaProcessor.py
• event/algorithm/processing/__init__.py
• test/event/test_tdoa_fdoa_processing.py

Files to Modify

• event/event.py - Integrate TDOA processing into main pipeline
• event/algorithm/localisation/ - Update localization to use TDOA measurements
• Configuration files - Add TDOA processing parameters

Benefits for MVP

• More accurate positioning with proper multi-static measurements
• Better handling of timing uncertainties
• Foundation for advanced localization algorithms
• Alignment with flowchart "Solver" component design

Integration with Existing System

Before RadarOnlyAssociator (#20)

Raw radar data → Need association first → Then TDOA processing

After RadarOnlyAssociator (#20)

Raw radar data → Association → TDOA processing → Localization → Tracking

Success Criteria

• Compute TDOA measurements from associated radar detections
• Handle clock synchronization issues appropriately
• Integrate with existing localization algorithms
• Improve positioning accuracy compared to direct delay processing
• Performance suitable for real-time operation
• Works with synthetic-adsb test data and real radar inputs

Dependencies

• Requires Implement RadarOnlyAssociator for pure radar tracking #20 (RadarOnlyAssociator) - Need associated detections first
• Should be completed before major localization improvements (Assess and improve localization accuracy (post-radar-association) #9)

MVP Priority

🟡 HIGH - Required for proper multi-static radar operation

[jehanazad]

I think this is accomplished by the TelemetrySolver and blah2 on the nodes. It accepts Doppler delay as the Bistatic range and solves based on this. The actual TDOA and FDOA are calculated by the nodes themselves, using their own single clock for both the reference and surveillance signals to avoid clock synchronization issues.