Engineer working on a laptop connected to a device test rig, surrounded by wiring harnesses and test hardware

Firmware vulnerabilities & fixes

ETSI TS 103 701 (conformance assessment for consumer IoT), explained

Last updated:

ETSI TS 103 701, currently V2.1.1 (published May 2025 by ETSI's Cyber Security (CYBER) technical committee, replacing V1.1.1 of August 2021), is the free companion Technical Specification that turns ETSI EN 303 645's 13 outcome-focused baseline provisions into a repeatable test methodology: concrete test groups, test cases, and pass/fail criteria that a lab, manufacturer or certification scheme can actually run against a device. Its own Introduction explains why a separate document exists at all: EN 303 645's provisions are "necessarily high-level and outcome-focused" because the standard covers an enormous range of consumer IoT device types, so TS 103 701 "seeks to contribute to a harmonised approach to assessing the conformance of consumer IoT products" against that baseline, independent of any specific certification scheme.

Key takeaways

  • The ETSI deliverables directory for TS 103 701 lists two published editions, and the document history in V2.1.1 records "V1.1.1 August 2021 Publication" and "V2.1.1 May 2025 Publication".
  • Implementation Conformance Statement (ICS): a provision-by-provision claim of what the DUT implements.
  • Phase 4 is easy to underestimate.
  • Test case 5.1-1-2 (functional): the TL then probes the device.

How deep does a TS 103 701 assessment go?

TS 103 701's own Scope clause is explicit about how deep the testing goes: the document "intends to contribute to the protection of consumer IoT products against the most common cybersecurity threats," but "multi-medium or highly targeted/sophisticated attacks and thus the invasive analysis of hard- and software modules is out of scope." The test scenarios "are targeting basic effort regarding test depth and test circumference in accordance to ETSI TS 103 645/ETSI EN 303 645 which addresses a baseline security level." The same clause adds that the document "does not describe specific tools or detailed step-by-step instructions"; the test cases are meant for "competent bodies that have the expertise to derive a suitable test plan." This matters for anyone citing a TS 103 701 assessment as evidence: it demonstrates conformance to a baseline, not resistance to a dedicated, well-resourced attacker. A "conformance assessed" claim should not be read as more than that.

What changed in V2.1.1 compared with V1.1.1?

The ETSI deliverables directory for TS 103 701 lists two published editions, and the document history in V2.1.1 records "V1.1.1 August 2021 Publication" and "V2.1.1 May 2025 Publication". The main differences a supplier will notice:

Topic V1.1.1 (2021-08) V2.1.1 (2025-05)
Normative baseline TS 103 645 V2.1.2 and EN 303 645 V2.1.1 (both 2020-06) TS 103 645 V3.1.1 (2024-01) and EN 303 645 V3.1.3 (2024-09)
Test groups / test cases (counted in clause 5 headings) 64 / 108 74 / 128, of which 72 conceptual
ICS pro forma in TS 103 701 (clause A.3) in Annex B of TS 103 645 / EN 303 645; TS 103 701 keeps the DUT identification (A.2) and IXIT (A.3) pro formas
"Reporting implementation" test group 4-1 5.0-1, following EN 303 645 V3's renumbering
Split provisions single test groups new A/B test groups, e.g. 5.1-2A, 5.3-4A/4B, 5.3-6A/6B, 5.3-15A/15B, 5.6-4A/4B, 5.13-1A/1B; data protection adds 6-6, 6-7 and 6-8

If a lab, customer or scheme asks for a TS 103 701 report, confirm which edition it expects. An ICS written against EN 303 645 V2.1.1 provision numbers does not map one-to-one onto the V3 provision list.

What are the three roles: DUT, SO, TL?

Role Who it is
Device Under Test (DUT) The specific consumer IoT device being assessed, "in live operation," tested via its offered interfaces with partial design knowledge from the IXIT (grey-box testing). Clause 4.2.1 says "the most up-to-date software version of the DUT should be used"
Supplier Organization (SO) "The developer, manufacturer, vendor or distributor of the DUT". It requests the assessment, completes the ICS and IXIT, and coordinates with component makers, service providers and app developers to do so
Test Laboratory (TL) Performs the assessment; clause 4.2.3 allows 1st party (a self-assessing part of the SO), 2nd party (a user organization) or 3rd party (an independent testing authority). Lab competence requirements such as EN ISO/IEC 17025 are explicitly out of the document's scope

What does the supplier have to produce: ICS and IXIT?

Before any testing starts, the SO completes two documents, defined in clauses 4.4 and 4.5:

  • Implementation Conformance Statement (ICS): a provision-by-provision claim of what the DUT implements. In V2.1.1 the pro forma is Table B.1 of EN 303 645 V3.1.3, which marks each provision M (mandatory), R (recommendation), C (conditional) or F (applies to a feature, capability or mechanism) and asks for Y, N or N/A plus a "Detail" entry. Mandatory provisions must be claimed; every "N/A" and every "N" needs a written justification.
  • Implementation eXtra Information for Testing (IXIT): the design detail that lets the TL plan and run tests. The SO only completes entries needed for provisions claimed "Yes"; Table B.1 of TS 103 701 maps provisions to IXIT entries, and clause 4.5 allows pointing to existing documentation instead, provided that documentation is handed to the TL. Annex C gives a complete sample IXIT for a fictional IP camera, "IPC 2000", that is worth reading before writing your own.

The V2.1.1 IXIT pro forma (clause A.3) has 29 numbered tables. The ones most device teams spend time on:

IXIT table What it records
1-AuthMech Every authentication mechanism, its authentication factor and the "Password Generation Mechanism"
4-Conf Yes/No confirmations, for example that vulnerability monitoring and update procedures are actually in place
5-VulnMon How vulnerabilities are monitored, identified and rectified
6-SoftComp Every software component (BIOS, firmware and boot loader are given as examples), its update mechanism, and whether it uses cryptography
7-UpdMech / 8-UpdProc Update mechanisms and the update procedure with its target "Time Frame"
10-SecParam / 11-SecComMech Security parameters and secure communication mechanisms
20-SecBoot Secure boot mechanisms
29-InpVal Input data validation methods per interface

The six-phase procedure in clause 4.3 runs: (1) the SO completes the DUT identification, (2) the SO completes the ICS, (3) the SO completes the IXIT, (4) the TL verifies the ICS, (5) the TL performs the test groups for every "Yes" claim, and (6) the TL assigns the overall verdict.

How does the lab check an ICS before testing?

Phase 4 is easy to underestimate. Clause 4.3 tells the TL to verify that:

  1. no mandatory provision is claimed "N";
  2. no conditional provision claimed "N/A" has an indication that its condition actually applies. The standard's own example: provision 5.3-8 claimed "N/A" when an update mechanism does exist is wrong;
  3. no feature-dependent provision claimed "N/A" relates to a feature the device actually has. Note 5 says a functional test case, such as 5.1-1-2 a), can be used to check for undeclared password-based login;
  4. "Yes" claims on conditional and feature-dependent provisions match the device;
  5. every "N/A" sits on a provision that is conditional or feature-dependent.

The TL can revise this check later if testing turns up contrary evidence, and an invalid ICS is on its own enough for an overall FAIL (Table 1).

How do conceptual and functional test cases differ?

Most test groups split into two test-case types (clause 4.1): conceptual test cases check "conformity of the IXIT against the requirements of the provision (conformity of design)," while functional test cases check the DUT, its associated services or its processes "against the requirements of the provision (conformity of implementation)." TSO 5.1, "no universal default passwords," shows the pattern:

  • Test case 5.1-1-1 (conceptual): the TL reviews the "Password Generation Mechanism" documented in IXIT 1-AuthMech for every password not defined by the user and checks on paper whether it "ensures that passwords are unique per device."
  • Test case 5.1-1-2 (functional): the TL then probes the device. It scans network interfaces for password-based authentication mechanisms the SO did not document (the test text notes "network scanning tools allow for discovery of network-based authentication mechanisms"), confirms user-defined passwords are required before use, and checks that real passwords do not contradict the documented generation method.

Test group 5.1-2 adds a password-quality check: whether a pre-installed password generation mechanism "induces obvious regularities" (its example is incremental counters like "password1", "password2") or draws from common password-dictionary strings, with a published top-100k common-password list as a reference. Some test groups are conceptual only. Test group 5.2-3, for example, assesses whether the documented way of monitoring, identifying and rectifying vulnerabilities is suited to the job, and checks that the IXIT 4-Conf confirmation is given.

How do PASS, FAIL and INCONCLUSIVE verdicts roll up?

Clause 4.6 builds verdicts bottom-up through three levels: test case, test group and overall.

Level PASS condition FAIL / INCONCLUSIVE
Test case (Table 3) Required elements present and the test case's own PASS criteria met INCONCLUSIVE when required elements, for example "evaluation tools and IXIT information", are missing or insufficient
Test group (Table 2) Every test case passes, or accepted external evidence under clause 4.7 FAIL if any test case fails; INCONCLUSIVE if one is inconclusive and none failed
Overall (Table 1) Valid ICS and a PASS for the test group of every "Yes" claim FAIL on an invalid ICS or any FAIL; otherwise INCONCLUSIVE if any group is inconclusive

Note 12 in clause 4.3 (Note 8 in V1.1.1) contains a practically important escape hatch: a FAIL or INCONCLUSIVE result tied only to a recommended provision does not block an overall PASS if the SO withdraws that provision's claim in the ICS with a justification, and the results already gathered can be reused. Clause 4.7 separately lets a TL accept external evidence, such as existing certifications or third-party evaluations of parts of the DUT, instead of running a test group. The TL must check that the evidence's scope, test activities and test depth match the test group; the SO flags the evidence in the ICS "Detail" column and hands over the certificates or reports.

What does a filled-in ICS and IXIT row look like?

Worked example, synthetic: a fictional Wi-Fi smart plug, firmware 2.4.0. It does not describe a real product.

Document Entry Content
ICS Provision 5.1-1 (M F) Y. Detail: per-device random password printed on the label; see AuthMech-1
ICS Provision 5.2-3 (R) Y. Detail: SBOM-based CVE monitoring per release; see VulnMon-1
ICS Provision 5.3-8 (M C) Y. Detail: OTA updates; see UpdMech-1
IXIT 6-SoftComp SoftComp-1 Linux kernel 5.10; update mechanism UpdMech-1; cryptographic usage: Yes
IXIT 6-SoftComp SoftComp-2 Boot loader; update mechanism: none, with the justification that it is immutable in ROM
IXIT 5-VulnMon VulnMon-1 Each release's CycloneDX SBOM is matched against NVD and CISA KEV; findings go to the PSIRT queue for triage and fix

Test group 5.3-1's objective notes that "any discovery of software components in the DUT is out of scope", so the TL works from the component list the SO gives. A complete, current SBOM is the practical way to make IXIT 6-SoftComp exhaustive, as clause 4.5 requires. See firmware SBOM for how to build one from a release image.

How does this fit with EN 303 645, PSTI and the CRA?

TS 103 701 only ever tests against EN 303 645's own provisions; V2.1.1 cites TS 103 645 V3.1.1 and EN 303 645 V3.1.3 as its normative baseline, and V1.1.1 cited EN 303 645 V2.1.1 of 2020. It is not itself named as a conformance route for the UK's PSTI regime or the EU Cyber Resilience Act in the text reviewed for this guide. Clause 4.8 leaves cryptographic requirements, tester expertise and accepted third-party evidence to whichever assessment scheme uses it. Where a scheme says it accepts a TS 103 701-based report, treat that scheme's current rules as authoritative. What TS 103 701 reliably gives a device maker is a structured, auditable assessment trail against EN 303 645, the same baseline UK PSTI draws its three mandatory requirements from (see the UK's own PSTI secondary legislation and our PSTI statement of compliance guide).

What is a workable order of work for a supplier?

  1. Pin the editionAgree with the lab or scheme whether the assessment uses V2.1.1 against EN 303 645 V3.1.3, and use the matching ICS pro forma.
  2. Build the ICS before touching the IXITClaiming provisions "Y", "N" or "N/A" first scopes which IXIT entries and test groups apply, per TS 103 701 Table B.1.
  3. Write the IXIT like a design spec, not a marketing summaryClause 4.5 requires "exhaustive and correct information", and an incomplete IXIT is explicit grounds for an INCONCLUSIVE verdict.
  4. Generate IXIT 6-SoftComp from the release SBOM, not from memory, and record cryptographic usage per component.
  5. Publish what the lab will check in live operation: the vulnerability disclosure policy, the support period and user information. Note 10 lets pre-release drafts be assessed first, with only the final publication checked later.
  6. Run the conceptual review first, functional tests second, as the methodology does.
  7. Decide 1st, 2nd or 3rd-party TL status deliberatelyA self-assessment is faster and cheaper but carries less weight with a customer or regulator than an independent assessment.

Which mistakes lead to INCONCLUSIVE or FAIL?

  • An undocumented login. A debug Telnet or SSH service found by the 5.1-1-2 scan, missing from IXIT 1-AuthMech, fails the test case.
  • "N/A" on a conditional provision that applies. Claiming 5.3-8 "N/A" while an OTA mechanism exists makes the ICS invalid, and an invalid ICS is an overall FAIL.
  • A thin IXIT. "We use industry-standard encryption" gives the TL nothing to test against; expect INCONCLUSIVE.
  • Testing an old build. Clause 4.2.1 expects the most up-to-date software version; results on an older build may not match what ships.
  • A missing justification for a recommendation. Test case 5.0-1-1 fails if any recommendation claimed "N" or "N/A" has no justification.
  • Mixing editions. V1.1.1 test group numbers in a V2.1.1 assessment confuse the mapping, especially for the split A/B provisions.

Does SoftWSP's automation help with the ICS/IXIT paperwork?

SoftWSP does not run TS 103 701's conceptual or functional test cases, does not act as a Test Laboratory, and does not analyse password generation, update mechanisms or cryptography. What its firmware pipeline does is turn a firmware image's software inventory into a CycloneDX 1.6 SBOM and match those components against NVD, CISA KEV, FIRST EPSS and CISA ICS advisories. An AI-drafted, cited report of the matches goes to an engineer for review. A Supplier Organization can use that component list as input to IXIT 6-SoftComp and the matching run as part of the procedure it describes in IXIT 5-VulnMon. Every ICS claim, IXIT entry and PASS/FAIL/INCONCLUSIVE verdict stays with the SO's engineers and the TL.

For consumer IoT device assessments that combine this firmware-level CVE and SBOM evidence with a review against the EN 303 645 baseline, see Device & Firmware Security.

This guide summarises the public ETSI TS 103 701 V2.1.1 (May 2025) and the earlier V1.1.1 (August 2021); it is not a substitute for the full text and does not constitute compliance or legal advice.

Put this guide to work

Frequently asked questions

What is ETSI TS 103 701?

'CYBER; Cyber Security for Consumer Internet of Things: Conformance Assessment of Baseline Requirements' — the ETSI Technical Specification that defines how to test and assess a consumer IoT product against the 13 provisions of ETSI TS 103 645 / ETSI EN 303 645. The current version is V2.1.1, dated May 2025 (it replaced V1.1.1 of August 2021), produced by ETSI Technical Committee Cyber Security (CYBER). Its own Introduction states the goal: 'The present document is independent from an assurance scheme and seeks to contribute to a harmonised approach to assessing the conformance of consumer IoT products against ETSI TS 103 645 [1]/ETSI EN 303 645 [2].' Scope is explicit about depth: it targets 'basic effort regarding test depth and test circumference' matching EN 303 645's own baseline security level, and 'multi-medium or highly targeted/sophisticated attacks and thus the invasive analysis of hard- and software modules' are out of scope.

What are the DUT, SO and TL roles, and what are the ICS and IXIT?

Three roles run through the whole methodology (clause 4.2): the Device Under Test (DUT) is the IoT product being assessed; the Supplier Organization (SO) — the developer, manufacturer, vendor or distributor — requests and supports the assessment; and the Test Laboratory (TL) performs it, which clause 4.2.3 notes can be 1st party (a self-assessing part of the SO), 2nd party (a user organization) or 3rd party (an independent testing authority). The SO produces two documents for the TL: the Implementation Conformance Statement (ICS), 'a statement... of the capabilities implemented in or supported by the DUT' (a Yes/No/N/A claim against each provision, with a required written justification for every No or N/A), and the Implementation eXtra Information for Testing (IXIT), which 'enable[s] the TL to perform appropriate test activities' and underpins the methodology's grey-box testing approach — the TL has partial design knowledge via the IXIT but is not assumed to have source-code-level access.

What is the difference between a conceptual and a functional test case?

Clause 4.1 splits most test groups into two aspects: conceptual test cases assess 'conformity of the IXIT against the requirements of the provision (conformity of design)' — checking what the SO says it built — while functional test cases assess 'conformity of the DUT functionality... against the requirements of the provision (conformity of implementation)' — checking what the device actually does. A worked example from Test group 5.1-1 (no universal default passwords): the conceptual Test case 5.1-1-1 has the TL check the IXIT's documented 'Password Generation Mechanism' description for whether it ensures per-device uniqueness; the functional Test case 5.1-1-2 then has the TL actually scan the DUT's network interfaces for undocumented password-based authentication mechanisms and verify real passwords match the documented generation method, with an explicit example that 'network scanning tools allow for discovery of network-based authentication mechanisms.'

How are PASS, FAIL and INCONCLUSIVE verdicts assigned?

Clause 4.6 defines verdicts bottom-up: a test group gets PASS only if every one of its test cases passes (or every applicable one does, with justified exceptions), FAIL if any test case fails, and INCONCLUSIVE if a test case result is inconclusive without any FAIL present. The overall verdict is PASS when, per Table 1, 'all selected test groups on the base of the claimed provisions in a valid ICS, at least all mandatory provisions, are fulfilled' — Note 12 of clause 4.3 (Note 8 in V1.1.1) clarifies that a FAIL or INCONCLUSIVE result tied only to a recommended (non-mandatory) provision does not block an overall PASS if the SO instead withdraws that provision's claim with a justification. Clause 4.7 additionally lets a TL accept 'external evidences' — existing documentation or prior test results — in place of re-running a test group from scratch, where the SO provides that evidence directly.

Sources

  1. ETSI TS 103 701 V2.1.1 (2025-05), Cyber Security for Consumer IoT: Conformance Assessment of Baseline Requirements (current edition; clauses 1-4, 5.0-5.3, A.3, Annex C, History)
  2. ETSI TS 103 701 V1.1.1 (2021-08), Cyber Security for Consumer IoT: Conformance Assessment of Baseline Requirements
  3. ETSI deliverables directory for TS 103 701 (version history)
  4. ETSI EN 303 645 V3.1.3 (2024-09), Cyber Security for Consumer Internet of Things: Baseline Requirements (the standard TS 103 701 assesses conformance against)
  5. UK legislation.gov.uk: The Product Security and Telecommunications Infrastructure (Security Requirements for Relevant Connectable Products) Regulations 2023
Was this guide helpful?

In this cluster

ETSI TS 103 701 (conformance assessment for consumer IoT), explained