VeriSpace

Tamper-proof evidence engine: pairs server-side raw GNSS spoof-checks & Galileo OSNMA ground signatures with automated Copernicus Sentinel-1/2 orbital diffs.

  • 0 Raised
  • 7 Views
  • 0 Judges

Categories

  • Challenge #3: Countering transnational crime and terrorism with space data

Description

💎 Idea

The Problem

In cross-border crime, hybrid warfare and environmental sabotage, field media is only as trustworthy as its metadata, and metadata is easy to fake. Investigators, border authorities, insurers and citizen reporters face two compounding threats:

  • GNSS spoofing & metadata tampering: Phone GPS fixes and EXIF timestamps can be spoofed, edited or replayed. GNSS interference is now routine near contested borders and critical infrastructure.
  • Generative AI & staged disinformation: Synthetic or recycled imagery makes it easy to fabricate an infrastructure strike or an industrial disaster, or to dismiss a real one as fake.

The Solution

VeriSpace is a two-factor Proof-of-Reality engine. It produces a signed evidence package that scores how strongly a piece of field media is tied to a real place, a real moment and a real physical change. It combines two independent sources:

  • On the ground: timing and navigation data authenticated by Galileo OSNMA.
  • From orbit: independent observation by Copernicus Sentinel-1 and Sentinel-2.

We never take a phone's reported GPS position on trust. We check it twice, independently.

Factor 1: Authenticated Ground Capture (seconds)
  • Commit-before-reveal. At capture, the app hashes the raw media, the sensor data (compass azimuth, IMU, camera intrinsics) and the Galileo E1-B I/NAV pages, including the OSNMA MACK tags for the current 30-second subframe. It signs this bundle with a hardware-backed device key and commits it to our server before the matching TESLA key is broadcast in the next subframe. Once the key arrives, we verify the tags with OSNMAlib. This binds the capture to one specific Galileo subframe. It cannot be backdated or assembled later from archived navigation data.
  • Independent position recomputation. The app uploads raw GNSS measurements (pseudoranges, Doppler, C/N0, AGC). Our server recomputes the position itself, using ephemeris authenticated by OSNMA, instead of trusting the coordinates the phone reports.
  • Spoofing and relay indicators. We flag the known signatures of spoofing and meaconing:
    • signal strength that is suspiciously uniform across satellites
    • sudden AGC jumps
    • clock or position discontinuities
    • Galileo disagreeing with other constellations
    • strong signals from satellites that the Copernicus DEM says should be hidden behind terrain
  • Legal timestamp. Each commitment is sealed with an eIDAS qualified electronic timestamp. This carries an EU-wide legal presumption that the time is accurate and the data hasn't changed.

Result: a Ground Trust Score with every check itemized, not a black-box yes/no.

Factor 2: Orbital Corroboration (next usable satellite pass)

Starting from the server-verified position and the camera's viewing cone, VeriSpace queries the Copernicus Data Space Ecosystem (CDSE). It pulls the last clean pre-event baseline and the first post-event Sentinel-2 and Sentinel-1 acquisitions, then computes change indices inside the cone:

  • ΔNDVI for clearing and smuggling tracks
  • ΔNBR for burn scars and explosions
  • Turbidity / NDWI for discharges
  • SAR backscatter change for excavation, flooding, large structural damage and oil slicks

Then it tests whether the view from orbit agrees with what the ground media claims. A fabricated "bridge collapse" over an intact bridge, or a "clearcut" over standing forest, fails here even if the photo itself is flawless.

Output: Evidence Dossier

A signed, C2PA-compatible package containing:

  • the original media hash
  • every Factor 1 check
  • the exact satellite scenes and change maps used
  • the scores
  • a plain-language methodology note

It is structured for chain-of-custody review in line with the Berkeley Protocol on Digital Open Source Investigations.

What we do and don't claim

Neither factor is a guarantee on its own:

  • OSNMA authenticates navigation data, not the radio signal path, so a sophisticated relay attack can still pass the cryptographic step. That is why we add measurement-level spoofing checks and orbital corroboration.
  • Sentinel resolution (10–20 m) verifies landscape-scale events, not individual objects.

We report calibrated confidence and state exactly what was and wasn't verified.

Business Case

B2B/B2G Verification-as-a-Service: a REST API and a mobile SDK (Android first), priced per verified dossier, with enterprise tiers. Three target markets:

  1. Defence, border security & law enforcement: EU and national agencies, environmental inspectorates and cross-border crime units verifying field reports.
  2. Catastrophe, agriculture & marine insurers: ground authentication within a minute at the moment of the claim, with orbital confirmation on the next usable pass. This cuts staged and recycled-photo fraud.
  3. Investigative journalism, NGOs & war-crimes investigators: cryptographically anchored chain of custody, plus independent satellite context.

Why VeriSpace

Existing evidence-capture apps such as eyeWitness to Atrocities already combine metadata with a secure chain of custody, and their footage has supported real prosecutions. VeriSpace adds two layers we haven't seen offered together: Galileo OSNMA authentication and automatic, independent Copernicus corroboration. We ship it as an SDK, so existing evidence platforms can embed it rather than compete with it.


🛰️ EU Space Technologies

Galileo OSNMA (Open Service Navigation Message Authentication)

  • What we use: OSNMA data in the E1-B I/NAV message: 40 bits per page (8-bit HKROOT + 32-bit MACK), 15 pages per 30-second subframe. The full chain of trust runs: Merkle tree root (from the GSC) → public key (DSM-PKR) → ECDSA-signed TESLA root key (DSM-KROOT) → TESLA chain keys → MAC tags authenticating ephemeris, clock and Galileo System Time parameters. Verification uses the open-source OSNMAlib.
  • Data sources: Android GnssNavigationMessage (TYPE_GAL_I) on supported devices. OSNMA-capable external receivers for drones and patrol vehicles.
  • Value brought: Authenticated ephemeris and time. These are the foundation for our server-side position recomputation and for the commit-before-reveal time binding. OSNMA has been operational since July 2025 and is free worldwide.
  • Roadmap: the Galileo Signal Authentication Service (SAS), which will authenticate the ranging signal itself and close the remaining relay gap.

Copernicus Sentinel-2 (MSI L2A)

  • What we use: 10 m bands (B2, B3, B4, B8) and 20 m bands (red-edge B5–B7, B8A, SWIR B11–B12), via the CDSE Sentinel Hub Process & Statistical APIs, with scene-classification cloud masking.
  • Value brought: automated before/after indices inside the camera's viewing cone:
    • ΔNDVI (10 m): illegal clearing and smuggling tracks
    • ΔNBR (20 m): arson, explosions and scorched infrastructure
    • NDWI, turbidity and red-edge chlorophyll proxies: illicit discharges into rivers and reservoirs

Copernicus Sentinel-1 (C-band SAR, GRD VV/VH)

  • What we use: dual-polarization backscatter, analyzed with log-ratio change detection through Sentinel Hub. Data comes from the current Sentinel-1C/1D constellation.
  • Value brought: works day and night, through cloud and smoke. Detects excavation, flooded breaches, large structural collapse, and oil slicks (dark VV patches on water). Criminals and saboteurs often operate at night or under cloud cover, and radar still sees them.
  • Roadmap: interferometric coherence change detection from Sentinel-1 SLC bursts (CDSE OData), processed in our own pipeline.

Copernicus DEM (GLO-30)

  • What we use: 30 m global elevation data.

  • Value brought: a terrain line-of-sight model around the recomputed position, used for two checks:

    1. Satellite visibility: which Galileo satellites terrain should block. A strong signal from a "blocked" satellite is a spoofing indicator.
    2. Skyline matching: the skyline the camera should see along its compass azimuth. A photo whose horizon doesn't match the DEM-rendered skyline lowers the score.

    The 10 m EEA-10 DEM is used where our partners qualify for access.


🛡️ EU Space for Peace & Resilience: Challenge Alignment

Primary challenge: #3, Countering transnational crime and terrorism with space data. Secondary: #1, Cybersecurity for space-based strategic enablers. VeriSpace hardens GNSS-dependent evidence workflows against spoofing and manipulation.

Countering Cross-Border Crime & Environmental Sabotage

Transnational networks exploit remote borders, forests and river basins for illegal logging, illicit waste dumping, smuggling-route excavation and sanctions evasion. VeriSpace turns a standard smartphone or patrol drone into a forensic sensor. Its reports are checked against Galileo cryptography and independent Copernicus observation.

Defeating Hybrid Disinformation & Spoofing

In contested regions where GNSS interference and deepfake propaganda are used to sow panic, VeriSpace flags capture-location anomalies (Factor 1) and staged or AI-generated "events" that contradict what satellites observed (Factor 2). It also does the reverse: it gives real evidence a defensible basis when someone dismisses it as fake.

Protecting Water Resources & Hydro-Infrastructure

In the Danube basin, Black Sea coastal corridors, dams and municipal reservoirs, VeriSpace verifies reports of illicit industrial discharge, bilge dumping, tailings-pond breaches and sabotage of hydro-infrastructure. It pairs authenticated ground evidence with Sentinel-2 turbidity and chlorophyll anomalies and Sentinel-1 surface-slick detection.


🤼 Team

Eduard Gurău | Team Lead, Full-Stack Architect & Systems Engineer CTO and software engineering founder specializing in high-throughput distributed backends (C# ASP.NET Core, Python FastAPI), cloud/container infrastructure, and interactive web/mobile interfaces (TypeScript, React Native, Next.js). Leads the VeriSpace core architecture, the commit-before-reveal pipeline and the Sentinel Hub orbital change-detection engine.

Matei Șerban | Full-Stack Engineer & Trust-Scoring Pipeline Developer Computer Science student at ASE Bucharest with hands-on experience building full-stack platforms in Next.js, React, TypeScript and PostgreSQL, alongside automated Python data pipelines using Scikit-learn and Pandas. He has engineered modular signal-extraction and custom scoring engines and handled B2B client relations. He leads VeriSpace's live verification interface and multi-factor trust-scoring engine.

Matei Sandu | Spatial Algorithms, Backend Engineer & Pitch Lead Computer Science student at ASE Bucharest (Merit Scholar, Top 9%), ISC2 Certified in Cybersecurity, and software developer skilled in Python, C# (.NET), C++, React and SQL. He draws on his Erasmus+ technical experience developing Python spatial optimization algorithms for dynamic object tracking, and on his background as a TED-Ed speaker. He engineers VeriSpace's spatial checks, including azimuth-cone geometry, DEM line-of-sight satellite visibility and skyline matching, and leads the stage pitch.

Vladimir-Cristian Sgondea | Cybersecurity, Compliance & Forensic Governance Lead Economic Informatics student at ASE Bucharest and Cybersecurity GRC Specialist working directly on EU NIS2 Directive compliance and the ISO/IEC 27001, 27002, 27005, 22301 and 27032 security standards. He combines practical Security-by-Design experience (data encryption, identity management, audit logging) with Python and AI training. He defines VeriSpace's chain of custody, eIDAS qualified timestamping, Berkeley Protocol alignment and European regulatory compliance.

Mihai Stelian | Space Data & Galileo/Copernicus Integration Specialist Computer Science & Economic Informatics student at ASE Bucharest focused on bridging GNSS telemetry and Earth Observation APIs with field-ready software workflows. He is responsible for raw GNSS measurement ingestion, the Galileo OSNMA verification layer (OSNMAlib / TESLA chain validation), and the automated Copernicus Sentinel-1 SAR and Sentinel-2 optical before-and-after queries.