Zebrid

Mission

Secure the future of orbital intelligence with autonomous, resilient and mission-grade data infrastructure.

Read the white paper

Self-healing data controller for orbital edge AI.

A radiation-resilient controller that sits between satellite sensors, storage, AI processors, and downlink — within a sub-payload power envelope.

See the controller
OREN-001 module — isometric wireframe CAD-style hairline drawing of the OREN data controller module showing PCB outline, central FPGA + RISC-V chip, two memory packages, and edge-connector pins. PCB multilayer OREN · FPGA + RISC-V [ OREN-001 ]
OREN-001 Controller card / SoM LEO Sub-payload envelope Status: HIL prototype

Challenge

Orbital data fails differently than terrestrial data.

In LEO, reliable data management is bounded by radiation, intermittent connectivity, power, and downlink — not by storage capacity.

01 · RAD Radiation → Silent corruption

Frequent SEUs.

Single- and multi-bit upsets flip stored bits with no warning. Trapped protons and heavy ions hit DDR, eMMC, and FPGA fabric at rates that scale with altitude and inclination.

02 · LINK Intermittent contact → Slow recovery

Multi-orbit gaps.

Ground-dependent fault recovery spans multiple orbits. A corruption detected at T+0 can wait an entire pass before an operator sees it.

03 · PWR Power & thermal → Scarce resources

Tight budget.

Storage recovery cannot consume the host compute envelope. Every watt spent on scrubbing is a watt not available to the payload's AI inference workload.

04 · DL Downlink limit → Manage on-board

Bandwidth-limited.

Downlink is bandwidth-constrained and expensive per bit. Sending corrupted or low-value data to the ground is wasted capacity on a finite, contended channel.

OREN Shield

A controller for the data path that feeds onboard AI.

OREN Shield verifies, repairs, and prioritizes data inside the satellite — before the AI processor ever sees it. Detection, ECC, scrubbing, replication, recovery, and link-aware scheduling, co-located with the data path, on a single radiation-resilient module.

Detect
Correct
Scrub
Replicate
Recover
Prioritize

Architecture

Sensor → Controller → AI inference → Downlink.

OREN data controller — eight-stage data path Horizontal hairline schematic. From left to right: payload sensors, ingestion, integrity verification, ECC and scrubbing, storage and replication, onboard AI processor, priority scheduling, downlink and ISL. The four middle stages — integrity verification, ECC scrubbing, storage replication, and priority scheduling — are highlighted as the OREN data controller. Payload sensors 01 · INPUT Ingestion 02 · I/O Integrity verification 03 · OREN ECC / scrubbing 04 · OREN Storage / replication 05 · OREN Onboard AI processor 06 · INFER Priority scheduling 07 · OREN Downlink / ISL 08 · LINK OREN DATA CONTROLLER · INTEGRITY + STORAGE OREN · SCHEDULER
Co-located with the data path FPGA / SoM acceleration Zero-copy DMA pipeline

Technical Notes

A white paper series on the OREN Project.

Zebrid is publishing a quarterly series of technical white papers covering the physics, architecture, and validation methodology behind OREN.

  1. N · 001

    Volume I: Industry Landscape & Quantitative Assessment

    Volume I establishes the foundational context, mapping the macroeconomic industry landscape, defining the core value propositions and application scenarios of data management, and formulating a unified evaluation matrix.

    • AUTHOR · DR. WEI MU
    Download PDF Read on arXiv
    PUBLISHING · 2026 Q2

Recognized by

Backed by investors and programs supporting the next generation of AI-native deep-tech ventures.