Low-Cost Space Situational Awareness Architecture

Executive Summary

Core Principle: Trade compute for aperture. Use software ingenuity and statistical methods to extract maximum information from minimum hardware cost.

Cost Target: 25× cheaper than traditional SSA system.

Approach	Cost	Capability
Traditional (1 professional site)	$330,000	Limited coverage
OpenASTRO (global distributed network)	$13,000	Full SSA pipeline

---

The Problem with Hardware-Heavy SSA

Traditional SSA requires: - Large aperture telescopes ($50,000-500,000 each) - Precise tracking mounts ($10,000-50,000 each) - Scientific-grade cameras ($10,000-30,000 each) - Dedicated facilities ($50,000-200,000) - Professional staff ($100,000+/year)

Total for 1 site: $330,000+

Problem: This doesn't scale. You need geographic distribution for triangulation.

Solution: Shift burden to software.

---

Software-Heavy Substitutions

1. Track-and-Stack → Large Aperture

Hardware approach: Buy 14" telescope to reach mag 15. Cost: $15,000-50,000

Software approach: Use cheap cameras + shift-and-stack.

Math:
  SNR improvement = √N (N = number of stacked frames)

  mag 11 camera + 100 frames → effective mag 13.5
  mag 11 camera + 1000 frames → effective mag 14.0

Implementation:
  1. Predict object motion from approximate orbit
  2. For each frame: shift pixels by predicted motion
  3. Accumulate shifted frames
  4. Result: Faint object becomes visible

Hardware saved: $15,000-50,000
Software cost: ~$0 (GPU already available)

2. Particle Filters → Precision Optics

Hardware approach: High-precision astrometry (0.5 arcsec) Cost: Premium site + mount + camera = $100,000+

Software approach: Massively parallel statistical sampling.

Traditional:
  1 perfect observation → Gaussian fit → orbit
  Problem: Expensive hardware

Software-heavy:
  100 noisy observations → 10,000 virtual particles → propagate → prune → orbit

Compute: Modern GPU handles 100,000 particles in seconds
Hardware saved: Entire precision optics budget

3. Distributed Network → Single Site

Hardware approach: Build one excellent site. Problem: Weather, equipment failure, single point of failure.

Software approach: Build many cheap sites.

Reliability math:
  P(all N sites down) = P_cloud^N

  For P_cloud = 0.5:
    N=1: 50% chance of NO observation
    N=10: 0.1% chance of NO observation

Cost for reliability:
  1 professional @ $50,000: 50% success rate
  10 cheap @ $1,500 total: 99.9% success rate

4. Edge Processing → High Bandwidth

Hardware approach: Stream video to central server. Problem: 25fps × 2MP × 50 sites = 2.5 Gbps

Software approach: Process on edge, upload summaries only.

Expensive:
  Raw video → central server → processing
  Bandwidth: 100 MB per detection
  Cost: $500+/month for bandwidth

Cheap:
  On-site RPi: detect streaks → extract positions → upload 1KB
  Bandwidth: 1 KB per detection
  Cost: $50/month (standard server)

Ratio: 1000× bandwidth reduction

---

Hardware Tiers (Cost-Obsessed)

Tier 0: Ultra-Cheap Detection Node ($100-200)

┌────────────────────────────────────────┐
│         DETECTION NODE                 │
├────────────────────────────────────────┤
│ Component          │ Cost  │ Notes     │
├────────────────────────────────────────┤
│ IMX307/IMX291 board│ $15-30│ Bulk      │
│ 8mm f/1.2 lens     │ $20   │ Fast glass│
│ Raspberry Pi 4     │ $55   │ Edge comp │
│ 3D printed case    │ $10   │ DIY       │
│ SD card + cables   │ $15   │ Standard  │
├────────────────────────────────────────┤
│ TOTAL              │$115-130│          │
└────────────────────────────────────────┘

Capability:
  - Detection limit: mag 10-11
  - Field of view: ~50° (wide)
  - Frame rate: 25-60 fps
  - Astrometry: 5-30 arcsec
  - Can detect LEO debris >30cm
  - Can detect GEO satellites

Why this works:

From Security cameras SSA.md: Sony STARVIS sensors (IMX291/IMX307) achieve: - Read noise: ~1.0 e⁻ (rivals scientific cameras) - Quantum efficiency: 80%+ (professional grade) - Frame rate: 60-120 fps (essential for streak detection)

These are mass-produced for security cameras, not astronomy. Economy of scale makes them 100-1000× cheaper than equivalent scientific cameras.

Tier 1: Entry Tracking Station ($500-1,000)

┌────────────────────────────────────────┐
│        TRACKING STATION                │
├────────────────────────────────────────┤
│ Component          │ Cost  │ Notes     │
├────────────────────────────────────────┤
│ 6" Dobsonian       │ $300-600│ Used    │
│ Alt-az mount       │ (included)│       │
│ Guide scope        │ $50    │ Finder   │
│ ZWO ASI290MM       │ $300   │ Or equiv │
│ Mini PC            │ $200   │ NUC/ equiv│
├────────────────────────────────────────┤
│ TOTAL              │$850-1,150│        │
└────────────────────────────────────────┘

OR (cheaper):

│ Modified webcam     │ $100   │   │
│ Used 6" Newtonian   │ $200   │ Craigslist│
│ Arduino controller  │ $30    │ DIY tracking│
├─────────────────────────────────────────┤
│ TOTAL               │$330    │       │

Why this works:

• 6" aperture reaches mag 13-14
• Doesn't need perfect tracking (software corrects)
• Used market has massive discounts
• Can track LEO with alt-az + software correction

Tier 2: Characterization Station ($0 - Partnerships)

┌────────────────────────────────────────┐
│   CHARACTERIZATION (NO OWN HARDWARE)  │
├────────────────────────────────────────┤
│                                        │
│ Option A: Data sharing agreements      │
│   - Amateur networks (free data share)│
│   - University partnerships (collab)  │
│   - Professional observatories (trade)│
│                                        │
│ Option B: Future: CubeSat deployment  │
│   - Low-cost lidar/photometric payload│
│   - Active sensing instead of passive │
│                                        │
│ Cost: $0 (leverage existing assets)   │
└────────────────────────────────────────┘

---

Network Topology

Minimum Viable: 3 Detection Nodes + 1 Tracking

        Node 1
       (North)
          │
          │
          │ 500-3000 km baseline
          │
Node 3 ───┼─── Node 2(West)      (East)
          │
          │
          │
      Tracking
       Station
       (Central)

Total: $450 (detection) + $330 (tracking) = $780

Capability: - Detection: Yes (wide fields) - Triangulation: Basic (altitude from 3 viewpoints) - Refinement: Limited (1 tracking station) - Characterization: No

Operational: 10 Detection + 3 Tracking

Total: $1,300 (detection) + $1,000 (tracking) = $2,300

Geographic spread: Continental
Coverage: Continuous (weather redundancy)
Triangulation: Good (multiple baselines)

Production: 50 Detection + 5 Tracking

Total: $6,500 (detection) + $1,700 (tracking) + 
       $1,000 (server) = $9,200

Geographic spread: Global
Coverage: Continuous, robust
Triangulation: Excellent
Refinement: Good
Characterization: Via partnerships

---

Software Components (All Free/Open Source)

Layer 0: Edge Detection

Runs on: Raspberry Pi 4 Language: Python + OpenCV Cost: $0 (open source)

# Pseudocode structure
class StreakDetector:
    def __init__(self):
        self.reference_frame = None
        self.star_positions = None

    def process_frame(self, frame):
        # Subtract static background (stars)
        # Find linear features
        # Hough transform for lines
        # Filter by velocity (distinguish from planes)
        # Return: (timestamp, x, y, angle, velocity)

Layer 1: Track Association

Runs on: Cloud VPS ($20/month) Language: Python + PostgreSQL + Redis Cost: $0 (open source)

# Pseudocode structure
class TrackAssociator:
    def associate(self, observations):
        # For each observation:
        #   Query known catalog for nearby objects
        #   If match: assign to existing object
        #   If no match: create new candidate
        # Link observations across time
        # Output: tracklets

Layer 2: Orbit Determination

Runs on: Cloud VPS (same server) Language: Python + Orekit Cost: $0 (open source)

# Pseudocode structure
class OrbitSolver:
    def determine_orbit(self, tracklet):
        if len(tracklet) >= 3 and arc_length > threshold:
            # Gauss method for good orbit
            return gauss_solution(tracklet)
        else:
            # Admissible Region for short arc
            return particle_filter_solution(tracklet)

Layer 3: Multi-Site Coordination

Runs on: Cloud VPS (same server) Language: Python + Celery Cost: $0 (open source)

# Pseudocode structure
class Scheduler:
    def schedule_simultaneous(self, target, sites):
        # Calculate visibility windows for each site
        # Find common window
        # Account for timing synchronization
        # Generate observation requests

---

Future: CubeSat Active Sensing

Why CubeSats?

Passive optical SSA has limits: - Only sees sun-illuminated objects (twilight hours) - Cannot measure range directly (needs triangulation) - Cannot see through clouds - Cannot see very small objects (<10cm)

Active sensing (lidar/radar) solves these, but ground-based radar is expensive ($100M+).

CubeSat advantage: Small, cheap, can be mass-produced.

Deployment Concept

┌─────────────────────────────────────────────────────────────┐
│             CUBESAT ACTIVE SENSING NETWORK                  │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│  CubeSat Swarm (10-50 units)                               │
│  ├── Each: 3U CubeSat ($50,000-100,000 to build)          │
│  ├── Payload options:                                      │
│  │   • Lidar: Range to debris, active illumination        │
│  │   • Flash imaging: High-resolution snapshots          │
│  │   • Radar: Bistatic with ground stations              │
│  └── Orbit: 400-500 km sun-synchronous                    │
│                                                             │
│  Ground Network:                                            │
│  └── OpenASTRO ultra-cheap nodes (already built)          │
│                                                             │
│  Total constellation: $5-10M                                │
│  vs. Traditional radar: $100M+                              │
│                                                             │
└─────────────────────────────────────────────────────────────┘

Lidar CubeSat Option

┌──────────────────────────────────────────┐
│        LIDAR CUBESAT PAYLOAD             │
├──────────────────────────────────────────┤
│ Component          │ Cost    │ Notes     │
├──────────────────────────────────────────┤
│ Laser (fiber)      │ $20,000 │ COTS      │
│ Detector array     │ $10,000 │ APD array │
│ Optics             │ $5,000  │ Custom    │
│ Spacecraft bus     │ $25,000 │ 3U standard│
│ Launch             │ $30,000 │ Rideshare │
├──────────────────────────────────────────┤
│ TOTAL PER UNIT     │$90,000  │          │
└──────────────────────────────────────────┘

Capability:
  - Range resolution: 1-10 m
  - Detection: >5cm debris
  - Day/night operation
  - No illumination requirement

---

Cost Comparison Summary

Ground-BasedOnly

Configuration	Detection Nodes	Tracking	Total Cost	Coverage
Minimal (MVP)	2	0	$260	Prototype only
Basic (triangulation)	3	0	$390	Regional
Operational	10	3	$2,300	Continental
Production	50	5	$9,200	Global

Ground + CubeSat (Future)

Configuration	Ground	CubeSats	Total	Capability
Enhanced	$9,200	$500,000 (5 units)	$510,000	Day/night, range, full 3D
Full	$9,200	$5,000,000 (50 units)	$5.1M	Professional-grade SSA

Compare to: Single professional site = $330,000

---

Key Metrics

Metric	Professional	OpenASTRO	Ratio
Cost per site	$150,000	$130	1150× cheaper
Sites for $50K	0.3	385	1283× more
Sky coverage	Limited point	Wide-field distributed	∞× better
Weather resilience	Single point	Multi-site redundancy	Robust
Astrometric precision	0.5 arcsec	5-30 arcsec	10-60× worse
Effective precision (N sites)	N/A	σ/√N ≈ 1.5 arcsec	Acceptable

---

Implementation Priorities

Phase 1: Minimal Detection Network (Months 1-3)

• [ ] Procure 3× IMX307 + RPi kits
• [ ] Deploy atgeographically separated sites (friends/colleagues)
• [ ] Implement edge detection software
• [ ] Set up cloud server for track association
• [ ] Validate with ISS observations

Phase 2: Tracking Refinement (Months 4-6)

• [ ] Add 1-2 entry tracking stations
• [ ] Implement orbit solver
• [ ] Test triangulation accuracy
• [ ] Correlate with TLE catalog

Phase 3: Scale Production (Months 7-12)

• [ ] Expand to 20+ detection nodes
• [ ] Add 3-5 tracking stations
• [ ] Implement full pipeline
• [ ] Begin catalog maintenance

Phase 4: CubeSat (Years 2-3)

• [ ] Design lidar/photometric payload
• [ ] Build and test CubeSat
• [ ] Launch rideshare
• [ ] Operate with ground network

---

References

• SSA.md (primary reference for science/tech details)
• Security cameras SSA.md (hardware specifications)
• Global Meteor Network (operational cheap sensor network)
• Project Luciole (fly's eye SSA architecture)