> ## Content Index
> Fetch the complete content index at: https://gridsignals.de/llms.txt
> Use this file to discover other available public pages before exploring further.

# Attest the Origin
- URL: https://gridsignals.de/attest-the-origin/
- Published: 2026-11-09T08:58:32.000Z
- Updated: 2026-11-09T08:58:32.000Z
- Description: Can grid operators trust the flexibility they buy? In Issue #013 the operator signed its answer. But the answer stands on data from devices that prove nothing about themselves. Here is how I would fix that, including for devices already installed.
- Author: Sascha Janssen

*Grid Signal — Issue #014 · November 2026*

---

**A signed answer, built on unsigned data**

In [Issue #013](https://gridsignals.de/proofs-not-data/), the transmission operator (TSO) asked a question and the distribution operator (DSO) sent back a signed answer: *"Together, at most 6 MW, valid until 14:15."* Later, anyone can check what the DSO said and when.

But the DSO's tools built that answer from data the devices send: the battery, the CHP units, the biogas plant. Each reports what it can do.

Now imagine one battery got a bad firmware update. It reports 3 MW of available power, but it can only deliver 1.5\. The DSO's tool believes it, the DSO signs, the TSO checks the signature. Every check passes. The answer is still wrong.

The TSO trusted a signature. The signature trusted a measurement. Nobody signed the measurement.

**So what?**

Three reasons this matters more every year.

*More of the grid depends on small devices.* Flexibility used to come from a few large power plants with staff on site. Now it comes from thousands of batteries, heat pumps and inverters that nobody visits. And EU rules now push distribution operators to buy flexibility, for example to manage congestion. The operator only sees what the devices report.

*Errors come in fleets, not one by one.* One firmware update goes to thousands of devices at once. If it is wrong, it is wrong everywhere at the same moment.

*Untrusted data has a price either way.* If the operator trusts every report, it pays for flexibility that may not exist. If it trusts none, it sets safe limits and blocks flexibility that does exist. Both cost money. Evidence makes the middle possible: its quality becomes an input to how much flexibility a DSO relies on, how tight it sets its operating limits, and how fast it can explain a gap between contracted and delivered flexibility.

**Why a certificate is not enough**

Today the grid answers "can I trust this device?" with certification: before a device may connect, its type is tested against the grid rules. But a certificate is a photo taken on day one. Five years later the device has new firmware, changed settings, maybe a replaced board. The photo still shows the model. It does not show this device today.

So there are two questions: was this type approved, and is this device running the software it should, right now? The certificate answers the first. A live signed answer depends on the second.

**What the wire can prove today**

How does the data travel from a device to the grid operator, and what can it prove on the way? It happens in two layers.

*Inside the site.* The inverter or battery talks to a local plant controller. The usual language here is Modbus, often in the SunSpec version. EPRI calls it "adopted almost universally by DER manufacturers". Modbus was built for closed factory networks. Basic Modbus has no security: anyone on the network can read or write values. A secure version over TLS exists, but it is far from universal.

*From the site to the grid operator.* Here the grid has stronger protocols. In Europe it is mostly IEC 60870-5-104, and IEC 61850 for protection and substations. In North America it is DNP3\. These protocols have security extensions: [IEC 62351](https://syc-se.iec.ch/deliveries/cybersecurity-guidelines/security-standards-and-best-practices/iec-62351/?ref=gridsignals.de) for the IEC protocols, and Secure Authentication for DNP3\. Depending on protocol and setup, they authenticate the sender and protect the message against changes.

That is the key point. Grid protocols can protect the message. None of them carries a proof of which software the device is running.

The same pattern shows up everywhere I looked:

- **United States.** The interconnection standard IEEE 1547-2018 asks that new distributed generation offers one of three interfaces: [SunSpec Modbus, DNP3 or IEEE 2030.5](https://restservice.epri.com/publicdownload/000000003002011591/0/Product?ref=gridsignals.de). California goes furthest: under [Rule 21](https://sunspec.org/ieee-2030-5-csip-certification/?ref=gridsignals.de), DER at the big utilities must by default use IEEE 2030.5, with certificates on every connection. But many inverters connect through a certified gateway or an aggregator's cloud, so the certificate identifies the gateway. And a certificate says who is talking, not which software runs.
- **Switzerland.** The Swiss rules are strict on the meter side. Under the [electricity supply ordinance](https://fedlex.data.admin.ch/filestore/fedlex.data.admin.ch/eli/cc/2008/226/20221001/de/pdf-a/fedlex-data-admin-ch-eli-cc-2008-226-20221001-de-pdf-a.pdf?ref=gridsignals.de) (StromVV, Art. 8b), smart metering systems may only be used if their parts passed a data security test by the federal metrology institute. A serious check, but done before use: a photograph again.
- **Middle East.** Dubai's utility DEWA spells out the layers very clearly in its [D33 solar specification](https://www.dewa.gov.ae/-/media/Images/d33/D33-Solar-PV-Monitoring-and-Control-Requirements.ashx?ref=gridsignals.de): inverters talk Modbus to a plant controller, the plant controller talks IEC 60870-5-104 to DEWA's control centres, and protection signals use IEC 61850\. The security requirements sit on the plant controller, for example separate authentication for remote firmware download. This protects the plant and its connection. It does not ask the inverter behind the controller to prove its state.

Three regions, three good approaches, one shared gap: everyone certifies the type or protects the channel. Nobody asks the device to prove what it is running when its data is used.

**How IT solved the same problem**

IT had this problem with routers and servers years ago. The answer is called remote attestation: the device proves which software and settings it started with. In simple terms, it works like this.

1. **A small security chip.** It holds a secret key that never leaves it. The best-known version is the TPM (Trusted Platform Module).
2. **A fingerprint at every start.** At boot, each step fingerprints the next piece of software before running it. The chip stores the fingerprints; later software cannot delete them.
3. **A fresh question.** A checker sends a random number. The chip signs the fingerprints together with it, and the checker compares them with known good values. The random number makes sure the proof is new, not a copy from last week.

The IETF describes the model in [RFC 9334](https://www.rfc-editor.org/rfc/rfc9334.html?ref=gridsignals.de), and [RFC 9683](https://www.rfc-editor.org/rfc/rfc9683.html?ref=gridsignals.de) describes it step by step for network devices.

For small devices that cannot afford a full chip, there is a lighter version called [DICE](https://trustedcomputinggroup.org/what-is-a-device-identifier-composition-engine-dice/?ref=gridsignals.de). It builds the device's identity from a hardware secret plus a fingerprint of its software. If someone changes the software, the device's identity changes too.

One important limit: attestation proves the measured software and configuration at a point in time, not that the device works correctly. A battery can run approved software and still have a faulty sensor or a worn power stage. Attestation does not replace measurements.

The technology exists. What is missing is using it in the grid.

**How I would solve it**

I would not wait for every device to be perfect. Four steps, each useful on its own.

![The chain from the device to the signed answer: devices, checker, gateway and meter feed evidence into the DSO grid tool, which sends a signed answer with its evidence mix to the TSO](https://gridsignals.de/content/images/2026/11/issue-014-diagram.png)

The chain from the device to the signed answer. Values and shares are illustrative.

*Step 1: Record the evidence for every reading.* Every value that goes into the DSO's tool carries what is known about it:

- **proven** — the device proved its software state with a security chip;
- **unchanged on the way** — a trusted box on site signed the value as it received it; the device itself proved nothing;
- **confirmed by the meter** — an independent measurement matches the reported outcome;
- **self-reported** — the device said so, nothing more.

These are not four levels of trust. They are different kinds of evidence, and one reading can have several: an attested device, a protected message, and a meter that confirms the result.

This costs little and works with today's fleet. The DSO's signed answer from #013 then carries the mix: *"6 MW, with attested devices behind 70 % of the available capacity; the other 30 % is self-reported."* The TSO knows how much weight the answer can carry.

*Step 2: Check when it matters, not once.* A device proof from last year is worth little. The DSO checks the device's state when it uses the reading for a decision, and a proof only counts as "proven" if it is recent. That is the same rule as in #013: answers expire, and so does evidence. One catch: a fresh proof of the software says nothing about a reading that did not come from that software. The proof and the reading must be tied together. That is a job for #015.

*Step 3: For old devices, rely on measurements, not on the gateway.* Most installed devices have no security chip and will stay connected for years. A gateway in front of them helps less than it seems. It proves its own state and that the report was not changed on the way. But if the battery reports 3 MW and can only deliver 1.5, the gateway signs the wrong 3 MW. Hence the label: "unchanged on the way". For old devices, the strongest independent evidence is usually the meter at the grid connection. But it measures the whole site. If the site also has solar panels and its own consumption, the meter sees the sum, not the battery. And it shows what was delivered, not what the battery could have delivered before dispatch. So the meter checks the site's actual exchange with the grid. Where a single device matters, you need its own measurement or a validated model of the site. A gateway adds real value only where it measures something itself.

*Step 4: Ask for proofs in new devices now.* Fleets turn over slowly, so start the clock today: operators can ask for remote attestation in their technical rules for new connections and flexibility products. Europe already points that way. The [Cyber Resilience Act](https://eur-lex.europa.eu/eli/reg/2024/2847/oj/eng?ref=gridsignals.de) applies to connected products sold from 11 December 2027\. Depending on the risk, it requires protection against unauthorised access and against changes to software and settings. It does not require a device to prove its state to a grid operator, and grid rules do not ask anyone to check. Product security and operational trust are two different things. New devices will have a lock. Somebody still has to check that the lock is closed when it matters.

And because no protocol on the way has a place for the proof today, step 4 also means a data path that can carry the proof next to the value. In #015 I will show what that could look like.

**Where I would start**

One DSO, one aggregator with a battery fleet, shadow mode:

- Record the evidence for every reading the DSO already receives. Nothing else changes.
- Compare what the sites report with what their connection meters measure. Count how often they disagree.
- Pick the next batch of new devices and switch on attestation where it exists.

After a few months you know two numbers: how much of your flexibility you can prove, and how often unproven reports were wrong. That is the business case.

**The chain, complete**

Issue #013 made the answer between operators checkable. This issue adds the link below it: the device proves its software, measurements check the outcome, and the signed answer says what evidence it stands on. Each link says what it checked, and nothing more.

In #015 I will build this chain step by step: what a device sends, what the checker verifies, and how the evidence ends up in the operator's answer.

Sign the answer. Then sign what the answer stands on.

---

**Sources**

- Grid Signal #013, Proofs, Not Data — https://gridsignals.de/proofs-not-data/
- IETF RFC 9334, Remote ATtestation procedureS (RATS) Architecture (January 2023) — https://www.rfc-editor.org/rfc/rfc9334.html
- IETF RFC 9683, Remote Integrity Verification of Network Devices Containing Trusted Platform Modules (December 2024) — https://www.rfc-editor.org/rfc/rfc9683.html
- Trusted Computing Group, Device Identifier Composition Engine (DICE) — https://trustedcomputinggroup.org/what-is-a-device-identifier-composition-engine-dice/
- Regulation (EU) 2024/2847, Cyber Resilience Act (Annex I; Art. 2 scope; Art. 71 application from 11 December 2027) — https://eur-lex.europa.eu/eli/reg/2024/2847/oj/eng
- IEC SyC Smart Energy, IEC 62351 security series — https://syc-se.iec.ch/deliveries/cybersecurity-guidelines/security-standards-and-best-practices/iec-62351/
- EPRI, IEEE Standard 1547 — Communications and Interoperability (three protocol options; Modbus "adopted almost universally by DER manufacturers") — https://restservice.epri.com/publicdownload/000000003002011591/0/Product
- SunSpec Alliance, IEEE 2030.5 / CSIP certification (California Rule 21) — https://sunspec.org/ieee-2030-5-csip-certification/
- Switzerland, Stromversorgungsverordnung (StromVV), Art. 8b Datensicherheitsprüfung — https://fedlex.data.admin.ch/filestore/fedlex.data.admin.ch/eli/cc/2008/226/20221001/de/pdf-a/fedlex-data-admin-ch-eli-cc-2008-226-20221001-de-pdf-a.pdf
- DEWA, D33 Solar PV Initiative Monitoring and Control Specifications — https://www.dewa.gov.ae/-/media/Images/d33/D33-Solar-PV-Monitoring-and-Control-Requirements.ashx