(1st Place for ANT61 USYD Hackathon)
Live Demo: https://hackathon-pied-eta.vercel.app/
Demo Credentials: test@example.com / password123
Executive Summary
System Overview
Core Capabilities
Real-Time Data Architecture
Technology Stack & Decision Making
Advanced Features
3D Visualization Engine
AI-Powered Threat Analysis
Real-Time Alert System
Database & API Integration
Performance & Scalability
Implementation Details
Future Roadmap
Technical Specifications
The Advanced Satellite Monitoring System is a web platform for satellite tracking and conjunction screening with an emphasis on space situational awareness. It ingests live data from CelesTrak, NASA, NOAA and Space-Track.org then fuses and serves it through an interactive command centre. Operators see a single view of active satellites and relevant debris with risk context and supporting analysis.
The platform runs continuous monitoring for collision risk and space weather effects. It presents results in a responsive 3D globe with alerting and a complete audit trail. The objective is to keep assets safe, enable timely decisions and record the rationale behind actions.
The system monitors satellites in real time, renders state in 3D and flags issues that need attention. It targets teams that manage fleets under tight timelines and cannot tolerate stale data.
Orbit density has increased. There are thousands of active satellites and a large catalogue of debris. This creates more conjunctions and higher operational load. The platform reduces that load by automating collection and first-pass analysis, then presenting clear actions to the operator.
The platform delivers live tracking from current TLEs with position, velocity and near-term predictions. It performs automated conjunction assessment with miss distance and probability estimates supported by encounter geometry. It overlays space weather events from NOAA and NASA on operational timelines. An ML layer assists with anomaly detection. A 3D interface turns orbital mechanics into an explorable scene. Alerts reach operators over the web, email and push so critical items do not wait.
Each function stands alone yet links to the others. An operator can move from an alert to the 3D view, to raw data, then to a report without leaving the application.
The service maintains a rolling picture of satellite state using live TLEs from CelesTrak and Space-Track.org. Positions update every 30 seconds. The tracker computes state vectors, short-term propagation and ground tracks for many objects in parallel. Redundant sources protect against temporary data gaps so the picture stays current.
A background service scans for close approaches, evaluates encounter geometry and estimates miss distance with relative speed. Items that cross risk thresholds are flagged. Each alert includes closest approach time, range and plane alignment with a confidence view. Operators can open a visualization of relative motion near time of closest approach to shorten analysis time.
Space weather can disrupt communications, power and attitude control. The platform ingests DONKI event feeds with NOAA indices, then maps expected impacts to affected assets and times. Operators can plan configuration changes or holds with this context visible in the same console.
WebGL rendering with Three.js and React Three Fiber provides a smooth globe with orbits, orbital planes and ground tracks. Users can scrub time, select objects, switch frames and inspect geometry. The scene is tuned for dense constellations and remains responsive under load.
The data path is designed for frequent updates without loss of integrity. Source adapters fetch, validate and normalise inputs. A processing layer enriches and fuses data, then publishes updates over WebSocket to subscribed clients. Changes are traceable to their origin.
The system integrates CelesTrak GP API for fresh TLEs and Space-Track.org for catalogues with conjunction data. It consumes NASA DONKI for solar events and NOAA Space Weather for geomagnetic indices with solar wind. Launch information from the SpaceX API is included where it improves conjunction context. N2YO supports pass prediction and visibility checks. Multiple feeds provide coverage and resilience so the operational picture remains usable if one source pauses.
Validation runs first with schema checks and unit conversions enforced by Zod. The pipeline deduplicates records and resolves conflicts by source priority with timestamp rules. Derived values such as ground tracks and propagation windows are computed before publishing. Outputs are versioned so any on-screen value can be traced to an input with time and source.
PostgreSQL on Supabase stores catalogues, conjunctions, alerts, weather events and analytics. Tables track satellites, orbital parameters, event timelines, alert acknowledgments and historical metrics. Real-time features push changes to clients without polling. Indexing and partitioning keep queries fast as data volume grows.
Next.js 15 with the App Router and Server Components keeps the client lean and first paint quick. Built-in code splitting and image handling reduce payloads. TypeScript support improves reliability. Deployment on Vercel provides global edges with automatic scaling.
Zustand supplies a compact store with strong TypeScript ergonomics. It avoids heavy boilerplate and fits real-time interaction patterns used in the console.
Zod enforces runtime schemas at the boundaries. It rejects malformed inputs early and stabilises UI state across rapid updates.
Three.js is the rendering core. React Three Fiber integrates scene updates with React so UI and visuals remain consistent. The stack is proven for complex dynamic scenes.
A dedicated WebSocket service broadcasts state changes and alerts. Socket.io manages connection lifecycle, rooms and fallbacks so operators remain connected across variable networks.
Models look for patterns that often precede anomalies such as drag spikes during storms or timing shifts after manoeuvres. The goal is early surfacing of likely issues. Models rank items by potential operational impact. Human operators make the final calls.
Alerts trigger when thresholds are crossed or when patterns match known risk. Each alert carries the context required to act with a link to the relevant visualization or data. Acknowledgment and escalation paths ensure that critical items do not linger.
Teams can review performance and incident timelines. Dashboards show alert volume, response time and recurring conjunction corridors. Reports export for audits and stakeholder updates.
The engine is optimised for clarity under load. It renders many objects smoothly and keeps interaction responsive.
The globe uses high-resolution base layers with a day-night terminator and atmospheric scattering. Boundaries remain readable at multiple scales. A polar view supports work with polar-orbiting assets. Visual treatment favours legibility over ornament.
Operators can view trajectories, planes and ground tracks with time controls. Constellation views reveal formation health and make phasing issues obvious.
Close approaches are drawn with relative motion and closest approach markers. Timers show proximity to TCA. Risk is colour-coded with labels for range and geometry. Where available, suggested avoidance vectors appear as overlays.
Camera presets, object selection, visibility toggles and measurement tools are available. Time can run live or be scrubbed. The layout supports fast pivots between the map, object panels and analysis views.
Models focus on anomaly detection rather than open-ended prediction. They flag deviations in drag, power or pointing that correlate with weather or manoeuvres. Training data comes from historical events with operator-verified labels.
Short-horizon forecasts indicate which assets are likely to enter higher-risk windows. This supports staffing plans and early configuration changes.
The system emits alerts for high-risk conjunctions and space weather events likely to affect operations. It also flags unexpected state changes and data quality issues. Severity levels map to notification channels and escalation rules so the right people are notified.
Alerts broadcast over WebSocket for instant delivery to all clients. Email and mobile push extend coverage off console. Unacknowledged items escalate by severity and age. Each alert maintains a complete history from creation to closure.
Core entities include Satellites, Conjunctions, Alerts, SpaceWeatherEvents, Manoeuvres and Analytics. Relationships link alerts to assets and events. Audit fields record acknowledgments with timestamps so actions are traceable.
REST endpoints expose satellite state, tracking, conjunction results, alert actions and analytics. WebSocket channels stream live updates. Resource shapes are consistent and strongly typed which simplifies integration with external tools.
Heavy computation runs off the UI thread. Server Components trim client JavaScript. Database partitions with tuned indexes keep reads fast. The 3D engine batches draws and reduces overdraw to stay smooth when scenes are dense.
Stateless services scale horizontally. WebSocket hubs shard by room to balance load. Background workers handle ingestion with queues so spikes do not affect the console. Storage and compute scale independently.
TypeScript is used end to end. ESLint and Prettier enforce style and correctness. Tests cover critical paths and data contracts. Continuous integration blocks merges on build and test status so main remains deployable.
Authentication and authorisation protect the console and the APIs. Roles follow least privilege. Transport is encrypted end to end. Secrets are stored in a managed vault. Logs are tamper-evident with retention aligned to policy.
Propagation can accept operator-supplied drag coefficients and areas during storm windows for higher fidelity. Manoeuvre planning will gain constraint-aware suggestions with operator approvals. Notes and playbooks will attach to alerts and assets to capture operational knowledge.
GPU-accelerated propagation will be evaluated for very dense screens. WebSocket rooms will gain finer granularity to limit chatter in large teams. Data provenance views will expand so source lineage is visible inline where decisions are made.
The console runs on modern browsers with WebGL support. A dedicated GPU improves 3D responsiveness. Server components deploy on Vercel with a managed Postgres instance. Real-time traffic is served by a separate WebSocket tier.
REST endpoints are versioned and documented. Authentication uses tokens tied to roles. WebSocket channels carry structured events with schemas enforced by Zod.
The reference deployment uses Vercel for the web app with edge functions, Supabase for Postgres and real-time features, and a containerised worker tier for ingestion with analytics. Infrastructure-as-code templates define environments and secrets to keep deployments consistent.
The Advanced Satellite Monitoring System gives operators a live and trustworthy view of their assets with the tools to act quickly. It fuses authoritative data, visualises orbital context and quantifies risk. Notifications reach the right people without delay. The architecture scales with fleet size and event volume, and the data model supports audits with collaboration.
Orbits are getting busier. Teams need clarity with speed. This platform focuses on both. It turns disparate feeds into an operational picture you can rely on and keeps the workflow tight from alert to resolution.
Built for today’s missions with room to grow.
Live Demo: https://hackathon-pied-eta.vercel.app/
Demo Credentials: test@example.com / password123