EM TwinThe Invisible Pillar of the Digital Twin
Backed by the dual foundation of Lauraycs ray tracing channel and hardware-in-the-loop (HIL), it builds a real-time, evolvable EM twin of the electromagnetic environment. It is both the infrastructure for end-to-end testing and the key link that feeds electromagnetic data back into city/factory/low-altitude digital twins—closing the RT loop with reality so it becomes new environmental data once again.
The Missing Link in the Digital Twin
Why the Digital Twin Lacks an "Electromagnetic Layer"
City, campus, factory, and low-altitude digital twin projects are emerging in droves, but the vast majority stop at the geometry and IoT-sensing layers. The moment wireless network planning, ISAC, low-altitude C2, satellite coordination, or intelligent-driving communication is involved, a missing "electromagnetic layer" means the twin cannot support end-to-end communication testing or strategic decision-making.
Geometry ≠ Electromagnetics
Existing digital twins only describe what is "visible," and are virtually blind to the reflection, diffraction, penetration, and time-varying multipath of electromagnetic waves.
Broken Reality Feedback
Coverage, interference, and packet-loss data from real networks cannot be fed back into simulation, leaving reverse calibration and optimization out of reach.
No Foundation for End-to-End
Without a continuously online electromagnetic environment that devices can access, end-to-end testing for autonomous driving / low-altitude / industrial use can only be fragmentary.
Solution
Turn the Electromagnetic World into a "Continuously Online" Twin Platform
The EM-Twin platform uses Lauraycs ray tracing as its electromagnetic kernel, overlays real-time data from live networks/factories/low-altitude/satellite-ground, and remains continuously visible to real terminals through its HIL subsystem. It is not a one-off simulation, but an electromagnetic environment service that can be subscribed to, queried, and driven.
Continuously Online
7×24 real-time EM field service; geographic scope, frequency band, and time are all subscribable; supports event-triggered replay.
Drivable
Directly drives hardware such as channel simulators, UAV C2 test benches, and satellite simulators via API/Streaming.
Data Closed Loop
Live-network/drive-test/factory operational data is fed back; measured calibration continuously improves twin fidelity, forming a bidirectional closed loop.
Platform Architecture
The Four-Layer Architecture of the EM Twin
From low-level environment data acquisition, to electromagnetic kernel computation, to outward-facing service interfaces, and on to integration with the upper-layer digital twin, it forms a complete EM twin platform. Every layer provides standard APIs.
Multi-Source Environment Data
OSM / BIM / oblique photography / laser point cloud / real-time IoT; aligned with city/factory digital twin base maps.
RT Electromagnetic Kernel
Lauraycs ray tracing continuously computes time-varying multipath and coverage; GPU clusters scale elastically.
Service-Oriented API
gRPC / REST / Streaming; subscribe to the electromagnetic state of a specified geographic box or time window.
Upper-Layer Applications and Feed Back
Network planning, ISAC, low-altitude, autonomous driving, HIL, AI training; real-world data is fed back for continuous calibration.
Platform Specifications
| Electromagnetic Kernel | Lauraycs cluster deployment; multi-GPU/multi-node parallelism; compute units partitioned by geographic box |
|---|---|
| Real-time Performance | Static coverage: second-level; dynamic multipath: ≤ 100 ms local refresh; triggerable millisecond-by-millisecond replay |
| Frequency Band | 600 MHz – 100 GHz (optional sub-THz extension) |
| Service Interfaces | gRPC, REST, WebSocket Streaming, Kafka data streams |
| Base Map Data | OSM / CityGML / IFC (BIM) / LAS point cloud / oblique photography; custom integration |
| Real-time Data Feed Back | Live-network KPIs (RSRP, SINR, RSRQ, CQI), drive-test PCAP, IoT sensing, vehicle/UAV trajectories |
| Digital Twin Integration | Already supports city-level, campus-level, factory-level, and low-altitude management (UTM) digital twin platforms |
| Visualization | 3D EM field visualization (coverage, shadowing, beams, multipath); timeline replay; event timeline |
| AI Training Data | Automatically produces labeled multipath/sensing/positioning datasets, usable for ISAC/AI-native algorithm training |
| Deployment | Private cloud / edge / self-controllable cluster (including ARM); supports tenant isolation and compute billing |
| Security | Data classification, access control, audit logs; supports MLPS and industry compliance |
| Self-Controllable | Kylin / domestic GPU adaptation; the self-controllable stack runs end-to-end |
Core Differentiation
Filling the electromagnetic gap
Truly "completes" the electromagnetic dimension of the digital twin, rather than statically pasting coverage maps as textures.
Drivable
Not just for viewing—it can serve as a channel source to drive HIL/OTA real-hardware testing.
Real-world closed loop
Measured data is continuously fed back, so twin fidelity converges over time rather than decaying.
Subscribable electromagnetic service
Allows third-party applications to subscribe to the electromagnetic state via API, with the twin serving as public infrastructure.
Cross-domain base maps
City, campus, factory, low-altitude, and satellite-ground twin systems can all be unified on a single electromagnetic kernel.
AI data factory
Outputs large-scale labeled datasets to support ISAC / 6G AI-native channel and positioning model training.
Typical Applications
Smart City
Coordinated planning of 5G private/public networks, electromagnetic situational rehearsal for major events, and emergency communication simulation.
Factory Digital Twin
Industrial 5G LAN / Wi-Fi 7 deployment optimization, communication link assurance for AGVs and collaborative robots.
Low-Altitude Management
UTM platform integration with EM-Twin, situational awareness for urban low-altitude C2/video links and route planning.
Integrated Satellite-Ground Network
Joint planning of LEO constellation coverage and ground shadowing, NTN service-level KPI prediction.
Intelligent Driving Road Network
End-to-end electromagnetic testing and strategy validation for roadside RSU/MEC and vehicle-side coordination.
AI / 6G Training
Large-scale, high-fidelity channel and sensing data for AI-native channel models and algorithm training.
Customers and Ecosystem
Let the digital twin see electromagnetics, and keep reality and simulation in a continuous closed loop.
乾径科技 MetaRadio · EM Twin Platform (EM-Twin)