What We Bring from Earlier Weeks

• Asset, threat, and the application protection plan — the written plan that governs a piece of software's security: scope, architecture, assets, threats, controls, verification, residual risk; where each asset lives, when it is created and destroyed, and which protection (C/I) it needs are listed in an asset table (Week 1)
• Independent evaluation, findings, and evidence — the evidence-based evaluation process run by an independent laboratory rather than the product's own developer; every gap the evaluator documents is a finding (Week 12)

This week: we link the asset table to every requirement, and we use the independent-evaluation discipline as a "requirement-to-evidence traceability chain" — the output is your project's compliance matrix (S17) and deferred requirements (S14).

What Is a Requirement?

• Requirement: a condition the system must meet.
• A security requirement: a security condition.
• A good requirement is verifiable (testable).

Criteria of a Good Requirement — Diagram

Three Types (Recap)

Criteria of a Good Requirement

• Verifiable: can it be tested?
• Singular: does it state a single thing?
• Measurable: is it concrete?
• Feasible: is it achievable?

Bad → Good (1)

• Bad: "The application must be secure."
• Good: "The release build must be produced with a stack canary, PIE, and full RELRO."

Why is it good? Testable (checksec).

Bad → Good Requirement — Diagram

Bad → Good (2)

• Bad: "Data must be encrypted."
• Good: "Every class-C asset must be encrypted at rest with AEAD at a level of ≥128 bits."

Why is it good? Which data, which level, which method is clear.

Bad → Good (3)

• Bad: "Keys must be well managed."
• Good: "(1) Every key's purpose, cryptoperiod, and destruction are documented. (2) It is erased from memory once its job is done."

Singular, verifiable items.

Worked Example · Improving a Bad Requirement

Starting sentence (from a real draft):

"The application must protect user data."

We will turn this sentence into a good requirement in six steps; at every step, before/after and why it changed.

Step 1 — Flag the Vague Words

• Before: "The application must protect user data."
• Vague words: "user data" (which data?), "must protect" (against what, at what level?).
• Why it changed: when an evaluator reads this sentence, they find nothing to test.

Step 2 — Link It to the Asset Table

• After: "Class-C fields in the local database must be protected."
• Why it changed: "user data" is not a single thing; it is separate rows in the asset table (Week 1) — session token, profile, payment token... each in a different class.
• "Must be protected" still isn't measurable → next step.

Step 3 — Clarify the Protection Goal

• After: "...the confidentiality and integrity of class-C fields must be protected."
• Why it changed: the threat model shows both "reading the file" (confidentiality) and "modifying the file" (integrity) as risks; both are needed at the same time.

Step 4 — Add a Measurable Technical Criterion

• After: "...must be protected with authenticated encryption (AEAD)."
• Why it changed: Week 9's rule — if confidentiality and integrity are both wanted at the same time, the right tool class is AEAD; still no specific library name is given.

Step 5 — Add the State and the Obligation Keyword

• After: "Sensitive class-C data at rest must be protected with authenticated encryption (AEAD)." (MUST)
• Why it changed: the data here is at rest (on disk); "must" was chosen because this is an indispensable condition against the threat — if it were "may," it would be treated as optional.

Step 6 — Give It an Id and Link It to the Threat

• After (final form): CEN429-DR-01 — "Sensitive class-C data at rest must be protected with authenticated encryption (AEAD)." (Threat: T-03, "an attacker who compromises the device reads/modifies the database file".)
• Why it changed: a requirement with no id never enters the traceability chain.

A Requirement's Life (animation)

A Requirement's Life — edge case: a one-word vagueness

Requirement Quality Checker (animation)

Requirement Quality Checker — edge case: a single sentence, a single rule

The Chain

Every requirement must be linked to evidence through this chain.

Why Is Traceability Important?

• The evaluator asks "where is this requirement?"
• Without traceability: searching everything from scratch.
• With it: found directly from the matrix.

A Traceability Example

Two Directions

• Forward: requirement → where was it met?
• Backward: this code/test → which requirement does it meet?
• A good matrix is traceable in both directions.

⚠️ A "Met" With No Evidence

• If the matrix says "met" but there is no evidence...
• In the evaluator's eyes it counts as not met.
• It becomes one of the first findings.

Worked Example · An End-to-End Chain

Let's fill in how the chain is built, start to finish, through a single asset: the user's session token.

Six links: asset → threat → requirement → control → verification → evidence.

1. Asset

2. Threat

• An attacker positioned on the network (e.g., on the same public Wi-Fi) can listen on an unencrypted connection and capture the token.
• Using the captured token in their own requests, they can act as the user.
• STRIDE: Spoofing + Information Disclosure (Week 1 terminology).

3. Requirement

The asset → threat → goal (confidentiality+identity) → tool class → state → obligation chain is briefly reapplied:

CEN429-DT-01 — "The session token must be carried over a channel using TLS 1.2 or higher, with the server certificate validated."

4. Control

The guide's S10.3 "Transport security" section:

"The client performs a TLS 1.2+ handshake on every connection to the server; the certificate chain and hostname are validated (see Week 10). No request is sent before the channel is established."

5. Verification

The security testing team tries to intercept the connection with a MITM (man-in-the-middle) tool:

• Verifies the connection is rejected when an invalid/self-signed certificate is presented.
• Verifies a plaintext (non-TLS) connection attempt is rejected.

6. Evidence

• The test tool's log output: mitm_test_2026-11-03.log
• CI run record: #617
• TLS handshake packet capture: handshake.pcapng
• Stored in the evidence/week13/ folder, referenced by these file names in the matrix.

Traceability Chain (animation)

Traceability Chain — edge case: every row a gap

Section 2's Rule

• Every "met" row's evidence column must have a findable, named reference: a file name, a CI number, a test report section.
• "Exists," "tested" is not evidence, it is a promise of evidence.
• The evaluator's most common finding: a "met" with no evidence.

Sections 1–2 — Quick Check

1. What are the four criteria of a good requirement?
2. Why is "the application must be secure" bad?
3. What is the traceability chain?

Sections 1–2 — Answers

1. Specific · measurable/verifiable · singular · traceable (feasible, atomic).
2. Vague and unmeasurable; "secure" has no definition → cannot be tested, everyone understands it differently.
3. Threat/standard → requirement → design/code → test/evidence. Every requirement is linked both ways to a source and to evidence.

Three Deferral Questions — Diagram

The Requirement-Block Pattern

Every requirement is written as a block:

• Id: a unique number (e.g., CEN429-DU-01)
• Text: what is asked for (singular, measurable)
• Status: met / deferred / not met
• Compliance: how it was met
• Verification + evidence

Example Block

CEN429-DU-01
Metin: Yerel DB'deki C sınıfı veri AEAD ile şifrelenir.
Durum: Karşılandı
Karşılama: AES-256-GCM, anahtar TEE'de
Doğrulama: T-05 birim testi
Kanıt: test çıktısı, S16

Why a Block?

• Every requirement has the same structure → comparable.
• The evaluator reads it fast.
• The compliance matrix is produced from these blocks.

What Is a Deferred Requirement?

• A component cannot meet a requirement itself.
• It defers the responsibility to the parent application/OS/hardware.
• This is not an escape, it is an explicit contract.

Three Questions When Deferring

Write, for every deferred requirement:

• To whom? (parent application / OS / hardware)
• Why? (why this component cannot meet it)
• How? (how the other party will meet it)

Deferred · Example

CEN429-AP-03: Güvenli kurulum ve güncelleme
Durum: Devredildi
Kime: Üst mobil uygulama (MPA)
Neden: SDK dağıtım kanalına sahip değil
Nasıl: MPA imzalı güncelleme + sürüm denetimi sağlar

⚠️ No Silent Deferring

• If you don't meet a requirement and don't write it down → a gap.
• Deferring must be documented; otherwise it counts as "not met."
• The S14 section exists exactly for this.

Met, or Deferred?

Worked Example · Filling In the Block (Met)

This is the written-into-the-guide form of the CEN429-DR-01 chain from Section 2:

[CEN429-DR] CEN429-DR-01 — MET
Requirement: Sensitive class-C data at rest must be protected with authenticated encryption (AEAD).
Compliance: The local vault file is encrypted with AES-256-GCM (S7.2); the key is derived from the device's secure storage unit. Verification: a unit test checks that decryption is rejected when a single byte changes. Evidence: CI #482, evidence/week13/test_butunluk_ci482.log.

The Same Pattern · Deferred

[CEN429-AP] CEN429-AP-07 — DEFERRED (to the parent application)
Requirement: The application must be installed and updated securely.
Compliance: The library has no distribution/update mechanism of its own. To whom: the parent application developer. Why: the library does not access installation at the network/file-system level. How: store signature validation, or update packages signed with the signing scheme from Week 6.

A deferred block also has a Compliance field — but it answers "how must the other party do it, why aren't we doing it" instead.

The Same Pattern · Not Met

[CEN429-DT] CEN429-DT-04 — NOT MET
Requirement: The server certificate must be pinned (certificate pinning).
Status: Currently only standard TLS chain validation exists, no pinning. Residual risk: MITM may be possible if a forged certificate is obtained from a trusted root (S16.4). Planned fix: v1.1 (task #217).

The "not met" block fills in the Status and Residual risk fields instead of Compliance.

The Role of the Block's Fields

• Header line ([Family] Id — Status): tells the evaluator's eye quickly where to go.
• Requirement line: repeats the "what" from the standard/threat, not reinterpreted.
• Compliance: describes the "how" with concrete file/section names.
• Verification + Evidence: the last two links of the traceability chain.

The Requirement Block (animation)

The Requirement Block — edge case: a silent hand-off

Section 3 — Quick Check

1. What are the requirement block's fields?
2. Which three questions are answered when deferring?
3. Why is silent deferring a problem?

Section 3 — Answers

1. Id · measurable statement · source/rationale · related asset/threat · verification method (evidence) · status/priority.
2. (1) Why isn't it met now? (2) What's the risk, who accepts it (approval)? (3) When/how will it be met (plan/mitigation)?
3. Without a record/approval, the requirement gets skipped → a hidden gap, no one owns it, it blows up in the audit. Defer openly, with a rationale, and approved.

Common Criteria — Diagram

What Is CC?

• Common Criteria: the international standard for evaluating product security.
• Provides a common language and method.
• Countries recognise each other's evaluations.

TOE · Target of Evaluation

• TOE (Target of Evaluation): the complete product being evaluated.
• Uniquely identified (Week 12): version + binary + hash.
• Out-of-scope components are written down.

ST · Security Target

• ST (Security Target): this product's security-target document.
• Contains: threats, objectives, requirements (SFR/SAR).
• Specific to the product.

PP · Protection Profile

• PP (Protection Profile): a common requirement set for a product class.
• Example: "mobile device PP," "smart card PP."
• STs generally build on a PP.

ST vs. PP

SFR · Functional Requirements

• SFR (Security Functional Requirements): what the product will do.
• Example: encryption, access control, logging.
• Chosen from a standard catalogue.

SAR · Assurance Requirements

• SAR (Security Assurance Requirements): how we will trust it was done correctly.
• Example: source review, test depth, vulnerability analysis.
• EAL determines these.

EAL · Evaluation Depth

• EAL (Evaluation Assurance Level): depth from 1 to 7.
• A higher EAL = a deeper review.
• "+" shows additional assurance components (EAL4+).

⚠️ A Common Misunderstanding of EAL

• EAL is the depth of security review, not the amount of security.
• Saying "EAL4+ > EAL2 is more secure" is wrong.
• Security depends on the threats and objectives in the ST.

Worked Example · ST Skeleton — TOE

Let's fill this in for a mobile payment component in three steps.

Step 1 — Define the TOE:

TOE: the "CEN429-Pay" library, version 1.0, Android ARM64 build only.
Out of scope: the parent application's interface, server-side components, the operating system kernel.

Writing what's out of scope matters as much as writing the scope.

The EAL Ladder (animation)

The EAL Ladder — edge case: EAL7, the whole ladder

FIPS 140-3 Levels — Diagram

What Is FIPS 140-3?

• The standard for validating cryptographic modules.
• The module's algorithms, self-test, key management.
• Covers only the module.

Security Levels (1–4)

FIPS 140-3 Levels (animation)

FIPS 140-3 Levels — edge case: Level 4

⚠️ The FIPS Compliance Trap

• Using a FIPS-validated library does not automatically make the application FIPS-compliant.
• The application must use the module in approved mode and correctly.
• It must also manage keys correctly.

CC and FIPS Together

• CC: evaluates the whole product.
• FIPS: validates the crypto module.
• A product can use a FIPS-validated module inside a CC evaluation.

Worked Example · A FIPS Level 1 Audit

The project's crypto layer: OpenSSL's EVP interface, AES-256-GCM, keys generated from getrandom(), and it silently falls back to a default key on error.

Let's check off the Level 1 expectations one by one.

Sections 4–5 — Quick Check

1. What is the difference between ST and PP?
2. What are SFR and SAR?
3. Why is "EAL4+ is always more secure" wrong?
4. Does a FIPS-validated library make the application FIPS-compliant?

Sections 4–5 — Answers

1. PP is a standard requirement set for a product class; ST is the document that says what a specific product (TOE) meets.
2. SFR is what the product will do (functional); SAR is how much it will be trusted (assurance; EAL comes from here).
3. EAL measures assurance depth, not the amount of security; if the scope (ST/TOE) is narrow, a high EAL covers little.
4. No. The module must be used in the correct mode/configuration, with approved algorithms. Module certificate ≠ application compliance.

Sector Standards — Diagram

Why Sector Standards?

• CC/FIPS are general; sectors want additional requirements suited to their own risks.
• IoT, mobile, payment are different threat environments.
• Follow whichever your project is closest to.

ETSI EN 303 645 (IoT)

• Baseline security requirements for consumer IoT.
• Example: no default passwords, secure updates, store sensitive parameters securely.
• A broad, applicable baseline.

ETSI · Example Provisions

• Unique passwords (no common default).
• Manage security updates.
• Securely store sensitive security parameters (→ project S5/S7/S8).

GSMA

• Security guidelines for the mobile operator ecosystem.
• Device, network, service security.
• SIM/eSIM, identity.

EMVCo

• Card payment ecosystem standards (cards, terminals, mobile payment).
• Functional conformance + security evaluation.
• Attack potential scoring (Week 12).

PCI

• PCI DSS: security for systems that process card data.
• PCI MPoC/CPoC: software-based payment acceptance.
• Strict; laboratory evaluation.

OWASP MASVS

• Mobile Application Security Verification Standard.
• Mobile application security requirements.
• Levels: baseline (L1), defence-in-depth (L2), resilience (R).

MASVS · For Your Project

• The most applicable requirement set.
• Tested with MASTG (Week 12).
• Crypto, storage, communication, code quality, resilience.

Comparison · The Lesson

• All share the same root: a written, verifiable requirement.
• The difference: scope and strictness.
• MASVS is the most practical starting point for your project.

Sector Standards Mapped Onto Our Families (animation)

Sector Standards Mapping — edge case: the family with no GSMA counterpart

Section 6 — Quick Check

1. Who is ETSI EN 303 645 for?
2. Why is MASVS the most suitable for your project?
3. What is the difference between CC and MASVS?

Section 6 — Answers

1. Baseline security for consumer IoT devices (no default passwords, updates, secure communication…).
2. The project is application-focused; MASVS has concrete, checkable items + the MASTG test guide → directly applicable.
3. CC is heavy, accredited lab/certificate (EAL); MASVS is a light, open, developer-friendly self-assessment standard.

Compliance Matrix Structure — Diagram

Requirement Carry-Over Decision — Diagram

Requirement → Asset → Control

Every requirement is linked to an asset and to a control.

The Link to Asset Management

• Requirements are linked to the S5 asset list.
• Every asset: C/I/I+ label, protection control.
• The requirement makes that protection mandatory.

Carrying It Into the Plan

• Functional requirement → design/code (S6–S11).
• Assurance requirement → test (S16).
• Process requirement → development process (S13).

Worked Example · Applying the Six Steps to CR-03

CEN429-CR-03 ("keys must be erased once their job is done"):

• 1. Applicability: there's a session key and a vault-file key → it applies.
• 2. Responsibility: the code is the project's own → we meet it ourselves.
• 3. Linking to assets: the vault-file key's "erasure" column is empty — the gap is noticed here.

The Same Example · From Threat to Release Plan

• 4. Linking to threats: links to the threat "an attacker with physical access takes a memory dump."
• 5. Control/verification: the crypto_wipe() call; verification: showing the key's bytes are not present in a memory dump.
• 6. Release plan: the session key was met in v0.9; the vault-file key is planned for v1.0, and until then written as not met + residual risk.

Carrying a Requirement Into the Plan (animation)

Carrying a Requirement Into the Plan — edge case: not applicable

Requirement Families — Diagram

The Course's Own Ids

This course uses its own requirement families, adapted from open standards:

CEN429-AP (application protection), ID (identity), AS (asset), DR/DU/DT (data: at rest/in use/in transit)…

Families (1)

Families (2)

Every Family → Links to Weeks

• AP → 4, 5, 6, 9, 11, 14
• CR/DR/DU/DT → 3, 10
• AS → 1, 3
• DV/RP → 12, 13

Your project chooses a subset of these families.

Compliance Matrix Activity · Step 1

• Choose five requirements from your own project (from different families).
• Write a requirement block for each one.

Activity · Step 2

Fill in a matrix row for every requirement:

Activity · Step 3

• Write at least one as deferred (to whom/why/how).
• Show at least one's evidence concretely.
• If there's a "not met," write it into the residual risk.

Activity · Evaluation

Check through an evaluator's eyes:

• Does every row have evidence?
• Are the deferred ones explicit?
• Are the requirements verifiable?

Fixing the Draft · Step by Step

1. Make the control concrete: a hash of the code section is computed at startup and compared with the embedded expected hash at build time; execution stops on mismatch (S12.3).
2. Clarify the verification method: a byte of the binary is changed, and it's verified the application refuses to start.
3. Add evidence: the CI job ci-integrity-check, run #391, integrity_test.log.

Where Is the Difference?

• The difference isn't in the row's length, it's in the concreteness of every cell.
• The first draft also looked "filled in," but no cell could be verified independently.
• What matters to the evaluator: being able to answer "how do we know this?" for every cell.

Section 8's Rule

• Every row's evidence must be specific to that row: a specific test name, a specific CI number, a specific log file.
• Copying a single general report into every row makes the whole matrix look suspicious.
• If referencing a general report, state which section/page of it too.

Sections 7–8 — Quick Check

1. Which two things is a requirement linked to?
2. Name three of the course's requirement families.
3. What must be in every row of a compliance matrix?

Sections 7–8 — Answers

1. Upward to a source (threat/standard/asset) and downward to evidence (test/code) — two directions.
2. (Any three of the nine families) e.g. data security · code hardening · RASP · crypto/certificates · memory protection · interface protection.
3. The requirement's id+statement · status (met/deferred) · evidence (test/file/S-section). A row with no evidence counts as not met.

What the Matrix Tells Us

• Three met (with evidence), one deferred, one residual risk.
• Every row is traceable.
• The evaluator starts from this table.

Project · S14 and S17

S17 and S14 are the documents an evaluator opens first.

• [ ] S17 compliance matrix: status, section, verification, evidence for every applicable requirement.
• [ ] S14 deferred requirements: to whom, why, how.
• [ ] If there's at least one "not met," write it into residual risk.

Through an Evaluator's Eyes

• A "met" with no evidence = not met.
• Are the deferred ones explicit and justified?
• Are the requirements verifiable?

Question 1

What are the four criteria of a good security requirement?

Answer: Verifiable, singular, measurable, feasible.

Question 2

Why is "the application must be secure" bad? How is it fixed?

Answer: It's unverifiable. Write it measurably: e.g., "the release build is produced with a stack canary, PIE, and full RELRO."

Question 3

What is the traceability chain?

Answer: Requirement → control → verification → evidence. Every requirement is linked to evidence.

Question 4

What does a "met" with no evidence mean in a compliance matrix?

Answer: It counts as not met for the evaluator; it's one of the first findings.

Question 5

Which three questions are answered when deferring a requirement?

Answer: To whom, why, how. Silent deferring counts as "not met."

Question 6

What is the difference between ST and PP?

Answer: ST is a single product's security target; PP is a common requirement set for a product class. ST generally builds on a PP.

Question 7

Is "EAL4+ is always more secure than EAL2" correct?

Answer: No. EAL is the depth of the evaluation; security depends on the threats and objectives in the ST. "+" is an additional assurance component.

Question 8

Is an application that uses a FIPS-validated library FIPS-compliant?

Answer: Not on its own. Validation covers the module; the application must use it in approved mode, correctly, and manage keys correctly.

Question 9

Where does ETSI EN 303 645's "securely store sensitive parameters" provision map to in the project?

Answer: S5 asset list, S7 data security/shell matrix, S8 key lifecycle.

Question 10

If a library cannot meet the "secure update" requirement?

Answer: It defers to the parent application and writes to-whom/why/how it will be met in the guide.