When an instrument lies, everything downstream inherits the lie
On 10 May 2025 a 304-metre container ship ran aground near Jeddah. The bridge team were not ignoring an alarm. They were reading a GPS receiver that showed a strong signal, a confident fix, and a position that was wrong. Two independent analysis firms concluded false satellite signals caused it. One of them called the attack unsophisticated.
Salvage took five weeks.
That is the problem SENTINEL exists for — and it is not a shipping problem. It is what happens anywhere a machine acts on several inputs and one of them is confidently wrong.
The gap nobody's alarm covers
Equipment failure looks like failure. It drifts, it drops out, it throws a fault code, and thirty years of alarm design will catch it.
An attack looks like normal operation. The signal is strong. The reading is plausible. Nothing is technically broken. No alarm fires, because by every measure the instrument is healthy — it is simply reporting something that is not true.
You do not catch that by watching one instrument harder.
What SENTINEL does
Compares. Every cycle it takes what each source says and evaluates them against one another, rather than trusting any single one — and rather than averaging a lie with the truth.
Counts. It tracks how many sources are genuinely participating. When interference removes one, the system knows it is running on fewer, instead of quietly carrying on.
Refuses. When the sources cannot agree, or too few are left, it says so and withholds the automatic action. It is built to fail closed, not to guess.
Proves. Every decision is sealed into a tamper-evident chained record, and the whole sequence replays exactly. Months later you can show precisely what the system saw and why it acted.
Why "proves" is the part that sells
Most assurance products tell you something went wrong. Very few can tell you, afterwards, what the machine knew at the moment it decided — which input arrived, which was refused and why, which never arrived at all, and what the system did with what was left.
An investigator, a regulator, an insurer and a court all ask the same question. SENTINEL is built so that question has an answer that is not a reconstruction.
Where it applies
| Maritime & defense | GPS jamming and spoofing at sea, assured position, and the record a casualty investigation needs |
| Aviation & autonomous systems | Aircraft, drones and self-driving: many sensors, one irreversible action |
| Critical infrastructure | Grid protection, substation timing, water and process control under a regulator's eye |
| Cybersecurity & AI agents | Operating safely while an input is already compromised, and consensus across AI systems |
Where this stands, stated plainly
- Four US provisional patent applications filed. Patent pending. None has been examined, and none is granted.
- A working reference implementation, with deterministic replay demonstrated.
- Not yet proven at sea or in the field. Results to date are from simulation. No sea trial, no hardware-in-the-loop, no installed base.
We would rather you read that list than discover it later.
Talk to us
RV Applied Systems, Inc. · Las Vegas, Nevada info@rvappliedsystems.com
We are looking for operators, insurers, integrators and researchers who want to test this against their own failure data.