IEC 62443 and Physical Access: What the Standard Really Asks
· 6 min read · Updated
Lire en français
The IEC 62443 standard, or more exactly the series of documents that carries this number, has become within a few years the reference that customers slip into their specifications and contracts as soon as industrial systems are involved, but it is dense, split into some fifteen documents aimed at different actors, and one often looks in vain for a direct answer to a simple question, which is what it requires regarding the person who physically walks up to a piece of equipment.
This article is for those who must meet an IEC 62443 requirement on an existing installed base, and who wonder what role physical access control plays in it.
A series in four families
Before going into detail, it helps to recall how the series is organised, because the answer to our question depends on which document one reads.
- Parts 1 set out concepts and vocabulary, that is zones, which group equipment sharing the same security requirements, conduits, which are the links between those zones, and security levels.
- Parts 2 address organisations, because part 2-1 describes the asset owner's security programme and part 2-4 the requirements for integration and maintenance service providers.
- Parts 3 describe the system, since 3-2 covers risk assessment and the division into zones and conduits, and 3-3 lists the system security requirements (SRs).
- Parts 4 address manufacturers, 4-1 for the development lifecycle and 4-2 for the technical requirements for components (CRs).
These requirements are grouped into seven foundational requirements, the same in 3-3 and 4-2, which are identification and authentication control (FR 1), use control (FR 2), system integrity (FR 3), data confidentiality (FR 4), restricted data flow (FR 5), timely response to events (FR 6) and resource availability (FR 7), and each can be met at four security levels, from SL 1 (casual or coincidental violation) to SL 4 (an attacker with extended resources).
What the IEC 62443 standard says, and where
To put it simply, the series was written by and for people who think in terms of communicating systems, because it describes very precisely what a component, network or application must do to identify a user, control their rights and record their actions, but it says much less about the very common situation where the user takes no digital path at all and simply presses a button.
Physical access is not absent, though, and in our reading it appears in three places, in three different ways.
In the asset owner's programme (2-1). The 2024 edition of part 2-1 reorganises the requirements into security programme elements and acknowledges, as we read it, that an installed base whose lifetime often exceeds twenty years cannot meet every technical requirement, so it leaves room for compensating measures defined by the asset owner, and that is where physical security of installations naturally belongs, alongside managing people's access.
In component protection (4-2). The component requirements include physical tamper resistance and detection for embedded devices, hosts and network devices, so the standard accepts that the attacker may have their hands on the hardware, but it treats the problem from the viewpoint of the component being opened, not of the machine being used without authorisation.
In compensating countermeasures. When a component cannot meet a requirement, which is the rule rather than the exception for a controller installed in 2003, the standard allows a compensating countermeasure to be provided by the surrounding system, and physical access control is one of the most natural answers to equipment that cannot identify the individual using it.
IEC 62443-3-3, requirement by requirement
It is in part 3-3 that the question becomes concrete, so take a piece of equipment with no user accounts, no log and no network, and look at what a control placed at its physical boundary can cover, and what it leaves aside.
| Requirement (IEC 62443-3-3) | Without physical control | With a control at the equipment boundary |
|---|---|---|
| SR 1.1 Human user identification and authentication | Impossible, because the equipment knows nobody | Compensation possible, since the person is identified before power-up |
| SR 2.1 Authorisation enforcement | Shared key or switch | Dated rule, with the right person, on the right equipment, at the right time |
| SR 2.8 Auditable events | None | Grants, denials and bypass attempts logged |
| SR 2.11 Timestamps | None | Every entry timestamped |
| SR 2.12 Non-repudiation | None | Signed entries, attributed to an identity |
| SR 6.1 Audit log accessibility | Not applicable | Logs retrievable and exportable to a SIEM, the tool that centralises security logs |
| SR 7.8 Control system component inventory | Declarative, often incomplete | Field inventory of protected equipment and of transmitters present |
| SR 3.x System integrity | Not addressed | Not addressed, because we do not touch the equipment's logic |
| SR 5.1 Network segmentation | Out of scope | Out of scope, since the equipment stays off the network |
Two observations seem important to us. The first is that the middle column is not a caricature, because it is the actual state of a large part of the installed base, and it is what, as far as we know, auditors who ask the question find. The second is that the right-hand column does not cover everything, and it would be misleading to claim otherwise, because a boundary control does not make the equipment more intact, but it makes its use attributable and verifiable.
What the auditor will ask for
In practice, an assessment against the IEC 62443 standard is not satisfied with a declaration of coverage, because it asks for evidence, and that is often where compensating measures fail, since an unevenly kept paper register proves nothing, and an unapplied procedure proves even less.
What gives a compensating measure its value is that it produces its own evidence, so a grant, a denial or a bypass attempt must leave a dated, attributed, tamper-evident trace without anyone having to remember to record it, and that is the property we sought by signing each log entry as it is produced, rather than relying on human discipline for traceability.
What we do not claim
The standard lends itself easily to marketing shortcuts, and we would rather be precise about our own position.
- Our architecture follows the design principles of IEC 62443-4-2, targeting security level SL-C 2, but no third-party assessment has been carried out to date, so we claim no certified compliance.
- No product makes a system IEC 62443 compliant, because compliance belongs to a system, assessed in its context, with its zones, conduits and security programme, and a product brings capabilities and evidence, which is already a lot, but it is not the same thing.
Where to start
If you must meet a requirement of the IEC 62443 standard on an existing installed base, we would suggest starting with a deliberately simple question, which is, for each critical piece of equipment in your zones, who can start it and how you would know tomorrow, because where the answer is "anyone" and "we would not", you have found the place where a physical compensating measure has the most value, and where it will be easiest to defend in front of an assessor.
For the layered model underlying the standard, our Purdue series starts with Purdue Diligence, and the link with the NIS2 directive is covered in How to Navigate NIS2 Compliance for Your Business.
Sources
- IEC 62443-2-1:2024, Security for industrial automation and control systems, Part 2-1: Security program requirements for IACS asset owners. IEC Webstore
- IEC 62443-2-4:2023, Security program requirements for IACS service providers. IEC Webstore
- IEC 62443-3-2:2020, Security risk assessment for system design.
- IEC 62443-3-3:2013, System security requirements and security levels.
- IEC 62443-4-2:2019, Technical security requirements for IACS components.
- ISA, overview of the ISA/IEC 62443 series. isa.org
