Real-Time Flight Visualization at Light Speed ✈️
HTTP/3 + WebGL + ADS-B
fusion architecture for live 3D air traffic
Current issues
High Latency
Traditional REST polling over TCP creates delays of >1 second
UI Blocking
Single-threaded processing freezes interface with thousands of aircraft
Limited Coverage
Local ADS-B receivers cover only 140-480km radius
Meet TAR Architecture
Evolution from tar1090 local UI to global real-time 3D visualization platform
HTTP/3 (QUIC)
UDP-based multiplexing eliminates head-of-line blocking
Web Workers
Parallel processing keeps UI responsive at 60 FPS
WebGL Overlay
GPU-accelerated 3D rendering on Google Maps
Hybrid Data Fusion
Local ADS-B
Sub-second latency from RTL-SDR receivers
OpenSky API
Worldwide coverage from volunteer network
Flightradar24
Rich verified data with flight details
HTTP/3 vs WebSocket
Why quick win?
0-RTT/1-RTT handshake
Stream-level multiplexing
Seamless network multiplexing
No packet loss blocking
Kalman Filter Fusion
1
Predict
Constant velocity model forecasts next state
2
Measure
Incoming data from multiple sources
3
Correct
Weighted fusion produces accurate position
3D Rendering Pipeline
01
Scene Init
Three.js scene synchronized with Google Maps 3D camera
02
Model Load
GLTF aircraft models with fallback geometries
03
Position Update
Lat/lon/altitude to 3D scene coordinates
04
Orientation
Heading, pitch, roll applied to model rotation
05
Render
GPU-accelerated draw at 60 FPS
Performance Results
<1s
Latency
End-to-end data to display
60
FPS
Stable frame rate under load
10K+
Aircraft
Concurrent rendering without drops
Future Roadmap
ML Trajectory Prediction
Forecast paths from state vectors and historical patterns
Anomaly Detection
Identify unusual behaviors and flight plan deviations
Trust Network
Community-driven data validation and quality scoring
Try OpenSky TAR Today
Download the APK or clone the repo to run your own instance
Star the repo if you're interested in real-time aviation visualization! 🚀
