Edge Computing, Enables the satellite to think!

OBC-Polar Vs. OBC-Polar-Edge

These edge computing solutions combine radiation-tolerant Microchip PolarFire FPGAs with NVIDIA Jetson Orin modules, delivering unprecedented on-board processing capability for next-generation satellite missions. This architecture enables satellites to move beyond traditional command-and-data handling and towards true on-orbit intelligence.

Edge computing refers to performing data processing, analysis, and decision-making directly on the satellite, rather than downlinking raw data to the ground for processing.

In space systems, this means:

Processing payload data immediately after acquisition
Extracting insights instead of transmitting large raw datasets
Making autonomous decisions in orbit with minimal ground intervention

This paradigm shift is increasingly important as satellite payloads generate higher data volumes and missions demand faster response times.

Edge computing OBCs are designed around a heterogeneous computing architecture:

PolarFire FPGA (Mission-Critical Layer)

Radiation-tolerant, low-power FPGA
Deterministic real-time control
Platform management, health monitoring, interfaces, and safety functions
Secure boot, fault tolerance, and watchdog supervision

NVIDIA Jetson Orin (High-Performance Compute Layer)

GPU-accelerated AI and machine-learning processing
High-throughput data handling
Support for modern AI frameworks (CUDA, TensorRT, ROS, PyTorch)
Optimized for computer vision, signal processing, and autonomy

This separation allows the FPGA-based OBC to remain in control of the spacecraft at all times, while the Jetson Orin operates as a powerful co-processor for payload intelligence.

How Edge Computing Differs from OBC-Cube-Polar

The OBC-Cube-Polar is a high-performance, flight-proven CubeSat OBC based on a PolarFire SoC FPGA with multi-core RISC-V processors. It is well suited for:

Command & data handling (C&DH)
Payload control
Moderate on-board data processing
Running RTOS or embedded Linux
Power-efficient, reliable CubeSat missions

However, edge computing platforms go significantly beyond this capability.

Key Differences

Aspect	OBC-Cube-Polar	CAVU Edge Computing OBC
Processing Type	Embedded RISC-V CPUs	FPGA + GPU-accelerated AI
AI / ML Capability	Limited / experimental	Native, high-performance
Data Processing	Moderate	Very high (real-time analytics)
Payload Intelligence	Basic	Advanced autonomy & inference
Data Downlink Strategy	Raw or lightly processed data	Event-driven, insight-based data
Mission Autonomy	Ground-centric	On-orbit decision making

In essence, OBC-Cube-Polar is an advanced spacecraft brain, while edge computing platforms add a powerful “cognitive layer” on top.

What Difference Does Edge Computing Make?

Reduced Downlink Bandwidth

Instead of transmitting full datasets (e.g. images, spectra, RF captures), the satellite can:

Downlink only relevant events
Send extracted features or alerts
Discard unimportant data autonomously

This dramatically reduces bandwidth requirements and ground station costs.

Real-Time On-Orbit Decision Making

Edge computing enables seeing, understanding, and acting in orbit:

Detect targets or anomalies immediately
Trigger follow-up observations
Adapt mission behavior without waiting for ground commands

This is critical for time-sensitive missions.

AI-Enabled Payloads

With NVIDIA Jetson Orin, missions can deploy:

On-board image classification and object detection
Change detection and monitoring
RF signal identification
Autonomous navigation and formation flying support

Increased Mission Autonomy

Edge computing shifts operations from ground-driven to spacecraft-driven, enabling:

Smarter constellations
Reduced operational workload
Faster response to dynamic environments

Mission Types That Benefit Most

CAVU’s edge computing OBCs are particularly well suited for:

Earth observation and surveillance
Hyperspectral and SAR payloads
Space situational awareness
RF intelligence and spectrum monitoring
Autonomous and responsive space missions
AI demonstration and in-orbit validation missions

For more traditional CubeSat missions focused on C&DH, power efficiency, and reliability, OBC-Cube-Polar remains an excellent solution.

CAVU Aerospace UK’s edge computing platforms represent the next step in satellite on-board computing. By combining radiation-tolerant PolarFire FPGAs with NVIDIA Jetson Orin AI processors, we enable satellites to process data where it matters most — in orbit.

While OBC-Cube-Polar delivers robust, high-performance spacecraft control, CAVU’s edge computing solutions unlock on-board intelligence, autonomy, and AI-driven missions, fundamentally changing what small satellites can achieve.

Understanding the Added Value of Edge Computing in Space

Architectural Comparison

Aspect	OBC-Cube-Polar	OBC-Polar-Edge
Core Architecture	PolarFire SoC FPGA (RISC-V CPUs)	PolarFire FPGA + NVIDIA Jetson Orin
Compute Model	Embedded, CPU-centric	Heterogeneous (FPGA + CPU + GPU)
Primary Role	Spacecraft brain (C&DH)	Spacecraft brain + AI payload processor
Autonomy Level	Limited / rule-based	Advanced, AI-driven
Mission Critical Control	Yes (primary)	Yes (handled by PolarFire FPGA)

Processing Capability

OBC-Cube-Polar

Multi-core RISC-V processors
Deterministic real-time performance
Suitable for:
- C&DH
- Payload control
- Data routing and compression
- Moderate on-board processing

OBC-Polar-Edge

Adds GPU-accelerated processing via Jetson Orin
Capable of:
- AI/ML inference
- Real-time image and signal processing
- Parallel compute workloads
Orders of magnitude higher processing throughput for payload data

Key Difference:
OBC-Cube-Polar processes data efficiently; OBC-Polar-Edge understands and interprets data on orbit.

AI & Machine Learning Capability

Feature	OBC-Cube-Polar	OBC-Polar-Edge
AI Framework Support	Experimental / limited	Native (CUDA, TensorRT, PyTorch, ROS)
GPU Acceleration	❌	✅
Neural Network Inference	❌	✅
On-Orbit Learning / Adaptation	❌	Limited / mission-specific

Edge Advantage:
Enables deploying terrestrial AI models directly in space, without off-loading computation to the ground.

Data Handling & Downlink Strategy

OBC-Cube-Polar

Collects and formats payload data
Relies on ground segment for:
- Interpretation
- Feature extraction
- Event detection
Typically downlinks raw or lightly processed data

OBC-Polar-Edge

Performs:
- Feature extraction
- Event detection
- Classification and filtering
Downlinks:
- Insights
- Alerts
- Reduced datasets

Impact:
Edge computing can reduce downlink volumes by 10×–100×, depending on mission profile.

Mission Autonomy

Capability	OBC-Cube-Polar	OBC-Polar-Edge
Rule-Based Autonomy	✅	✅
AI-Driven Decisions	❌	✅
Adaptive Mission Planning	❌	✅
Event-Triggered Actions	Limited	Advanced

Edge Advantage:
Satellites can observe, decide, and act without waiting for ground commands.

Interface & Payload Support

Both platforms support standard spacecraft interfaces (SpaceWire, CAN, RS-422, SPI, I²C), but:

OBC-Cube-Polar is optimized for traditional spacecraft subsystems.
OBC-Polar-Edge adds high-bandwidth data paths suitable for:
- High-resolution imagers
- Hyperspectral sensors
- RF digitizers
- Vision-based payloads

Power & System Considerations

Parameter	OBC-Cube-Polar	OBC-Polar-Edge
Power Consumption	Low to moderate	Higher (AI compute dependent)
Thermal Design	Simpler	More demanding
System Complexity	Lower	Higher

Trade-off:
Edge computing requires more power and thermal management, but delivers significantly higher mission value.

OBC-Cube-Polar – Best Fit For

Traditional CubeSat missions
C&DH-centric spacecraft
Power-constrained platforms
Missions with strong ground-segment processing
Technology demonstration & institutional missions

OBC-Polar-Edge – Best Fit For

Earth observation & surveillance
AI-enabled payloads
Responsive and autonomous missions
High-data-rate sensors
Constellations requiring local intelligence
Commercial missions where data value > data volume

OBC-Cube-Polar is a high-performance, reliable spacecraft OBC.
OBC-Polar-Edge builds on this foundation by adding on-orbit intelligence, enabling satellites to process, understand, and act on data in space.

In short:

OBC-Cube-Polar controls the satellite.
OBC-Polar-Edge enables the satellite to think.