1.1: SYS.1 Requirements Elicitation


What You'll Learn

Here's what you'll take away from this section:

  • Describe the purpose and outcomes of SYS.1
  • Apply base practices for requirements elicitation
  • Integrate AI assistance appropriately (L0-L1)
  • Produce compliant stakeholder requirements documentation

Process Definition

Purpose

SYS.1 Purpose: To gather, process, and track stakeholder requirements.

Outcomes

Outcome Description
O1 Stakeholder requirements are specified
O2 Consistency and bidirectional traceability are established between stakeholder requirements and the relevant input sources
O3 Prioritized stakeholder requirements are agreed with the stakeholders
O4 The agreed stakeholder requirements are communicated to all affected parties

Base Practices with AI Integration

BP Base Practice AI Level AI Application
BP1 Specify stakeholder requirements L1 Requirements drafting, template assistance
BP2 Ensure consistency and establish bidirectional traceability L2 Automated trace linking
BP3 Agree on stakeholder requirements L0 Human-only (stakeholder negotiation)
BP4 Communicate agreed stakeholder requirements L1 Documentation generation

BP Detail

BP1: Identify Stakeholders

Activities:

  • Map all parties with interest in the system
  • Document stakeholder roles and responsibilities
  • Identify stakeholder representatives

Stakeholder Categories:

Category Examples
End users Operators, maintenance personnel
Customers OEMs, tier-1 suppliers
Regulatory Safety authorities, standards bodies
Internal Development, manufacturing, service
External Suppliers, partners

AI Assistance: None (L0) - Human interpersonal judgment required

BP2: Obtain Stakeholder Requirements

Activities:

  • Conduct interviews, workshops, surveys
  • Review existing documentation
  • Analyze similar systems
  • Capture expectations and constraints

Techniques:

Technique When to Use
Interviews Deep understanding of individual needs
Workshops Consensus building, conflict resolution
Document review Existing systems, standards
Prototyping UI/UX requirements
Observation Operational context understanding

AI Assistance (L1):

  • Pre-interview analysis of existing documents
  • Summarization of previous meeting notes
  • Draft question generation

HITL Pattern: Collaborator - AI helps prepare, human conducts

BP3: Document Stakeholder Requirements

Activities:

  • Write clear, unambiguous requirements
  • Use consistent format and terminology
  • Capture rationale and constraints
  • Include acceptance criteria

Requirement Attributes:

Attribute Content
ID Unique identifier
Description Clear statement of need
Rationale Why this requirement exists
Source Which stakeholder
Priority Business priority
Acceptance How to verify satisfaction

AI Assistance (L1):

  • Draft requirement text from notes
  • Suggest consistent terminology
  • Identify ambiguous language

HITL Pattern: Reviewer - AI drafts, human reviews/approves

BP4: Analyze Requirements for Completeness

Activities:

  • Check for missing areas
  • Identify conflicts and overlaps
  • Verify clarity and testability
  • Assess feasibility

Completeness Checklist:

Area Covered
Functional requirements Yes/No
Performance requirements Yes/No
Interface requirements Yes/No
Environmental constraints Yes/No
Safety requirements Yes/No
Security requirements Yes/No
Regulatory requirements Yes/No
Operational requirements Yes/No

AI Assistance (L1):

  • Automated completeness checking
  • Conflict detection
  • Ambiguity flagging

HITL Pattern: Reviewer - AI flags issues, human resolves

BP5: Agree on Requirements with Stakeholders

Activities:

  • Present requirements for review
  • Resolve conflicts and ambiguities
  • Obtain formal acceptance
  • Document agreements

AI Assistance: None (L0) - Human interpersonal judgment required

BP6: Establish Traceability

Activities:

  • Link requirements to sources
  • Prepare for downstream tracing
  • Enable impact analysis

Traceability Links:

The diagram below illustrates how elicited stakeholder requirements link to their sources and downstream system requirements, enabling end-to-end traceability.

Requirements Elicitation Flow

AI Assistance (L1-L2):

  • Suggest trace links
  • Validate trace completeness
  • Identify orphan requirements

HITL Pattern: Monitor - AI maintains, human verifies


Work Products

Input Work Products

WP Description Source
Customer docs RFQ, contracts, specs Customer
Standards Regulatory requirements External
Previous systems Lessons learned Organization

Output Work Products

WP ID Work Product Content
17-03 Stakeholder requirements Documented needs and expectations
15-01 Meeting minutes Elicitation session records
17-11 Traceability record Links to sources

Example: Automotive ECU Stakeholder Requirement

---
ID: STK-BCM-001
Title: Door Lock Response Time
Source: Vehicle Manufacturer (OEM)
Priority: High
---

## Description

The body control module shall lock all vehicle doors within 200ms
of receiving a lock command from the central locking button.

## Rationale

Driver expects immediate response to locking action for
security confidence and feedback.

## Constraints

- Response time measured from button electrical signal
- All doors include front left, front right, rear left, rear right
- Trunk lock is handled separately

## Acceptance Criteria

1. Lock command to door actuator signal: < 150ms
2. Door actuator to locked state: < 50ms
3. Verified under temperature range -40°C to +85°C

## Trace

- Source: OEM Requirements Spec v2.1, Section 4.3.1
- Derives: SYS-BCM-010 (system requirement)

AI Integration Examples

L0 Activities (No AI)

  • Stakeholder identification
  • Interview conduct
  • Requirement agreement
  • Conflict resolution

L1 Activities (AI Assists)

Example: Pre-Interview Document Analysis

Input: Previous meeting notes, customer specifications

AI Action: "I've analyzed the customer specification and identified
           these potential requirement areas that weren't discussed
           in previous meetings:
           1. Cybersecurity requirements (Section 7)
           2. OTA update support (mentioned briefly in Section 12)
           3. Power consumption limits (Table 4-3)"

Human Action: Add these topics to interview agenda

L2 Activities (AI Generates)

Example: Traceability Suggestion

AI Action: "Based on requirement text analysis, I suggest these
           trace links:
           STK-BCM-001 ──► SYS-BCM-010 (99% confidence)
           STK-BCM-001 ──► SYS-BCM-011 (87% confidence)
           STK-BCM-002 ──► [No match found]"

Human Action: Confirm 99% link, evaluate 87%, investigate orphan

Common Challenges

Challenge Mitigation
Incomplete stakeholder identification Systematic stakeholder mapping
Ambiguous requirements AI language analysis + human review
Conflicting requirements Explicit conflict resolution process
Missing non-functional requirements Comprehensive checklist
Evolving requirements Change management process
Conflicting stakeholder requirements Explicit prioritization using MoSCoW

Requirements Elicitation Techniques

Overview of Elicitation Methods

Effective requirements gathering requires multiple complementary techniques:

Technique Best For Effort Accuracy Use Cases
Interviews Understanding needs, context Medium High Stakeholder discovery, domain knowledge
Workshops Group decision-making, consensus High High Trade-off discussions, conflict resolution
Prototyping Validation, user interface feedback High High Novel features, user experience validation
Observation Actual user behavior, work processes Medium High Understanding actual vs. stated needs
Surveys/Questionnaires Quantitative feedback, scalability Low-Medium Medium Broad stakeholder feedback, prioritization
Document Analysis Regulatory, existing systems Low Medium Standards compliance, legacy system understanding
Simulation/Mockups Feasibility, visualization Medium Medium Complex workflows, system behavior
Focus Groups Consensus building, feedback Medium Medium User segments, market validation

Elicitation Technique Details

1. Interviews

Preparation:

  • Create interview guide (not a script - allow flexibility)
  • Identify 5-10 key stakeholders per category
  • Schedule 1 hour per interview
  • Prepare opening, probing, and closing questions

Interview Structure:

Opening (5 min):
├─ Introduce purpose of interview
├─ Explain confidentiality/use of information
└─ Ask permission to record/take notes

Main Questions (40 min):
├─ Open-ended: "What are the main challenges you face?"
├─ Probing: "Can you give me an example?"
├─ Validating: "So you mean... is that correct?"
└─ Non-leading: Avoid "Don't you think...?" style questions

Closing (10 min):
├─ Summarize key points captured
├─ Ask "Did we miss anything important?"
└─ Thank and confirm next steps

Example Questions (Automotive Infotainment System):

  • "What are the top 3 pain points with current infotainment?"
  • "How often do you use voice commands versus touchscreen?"
  • "What would make the system more intuitive?"
  • "Are there safety concerns with any features?"

Documentation:

  • Audio recording (with permission) + summary notes
  • Direct quotes for key requirements
  • Action items and follow-up questions

AI Assistance:

  • L1: Auto-transcription of interviews (whisper, Azure Speech)
  • L1: Draft requirement extraction from transcripts
  • HITL: Human validates extracted requirements against intent

2. Workshops (Requirements Workshops)

Preparation:

  • 6-12 participants (representatives from all stakeholder groups)
  • Half-day session (4 hours optimal)
  • Facilitator + scribe + SME
  • Prepare agenda, materials, room setup

Workshop Agenda:

Opening (15 min):
├─ Overview of project
├─ Goals for workshop
└─ Ground rules (respect, listen, build on ideas)

Part 1: Individual Brainstorming (30 min):
├─ Silent generation: Each person writes requirements
└─ No criticism, no evaluation

Part 2: Consolidation (60 min):
├─ Group discussion of ideas
├─ Identify common themes
├─ Resolve conflicts (voting, structured discussion)

Part 3: Validation (45 min):
├─ Review consolidated requirements
├─ Assign priorities (MoSCoW)
└─ Identify gaps

Closing (30 min):
├─ Summarize outcomes
├─ Confirm next steps
└─ Thank participants

Conflict Resolution in Workshops:

The following diagram depicts a prototype AV system status display, used during stakeholder workshops to elicit concrete feedback on information layout and priority.

AV System Status Display

Feedback from Users:

  • "I want a confidence indicator (% chance of failure)"
  • "Show battery/range like my phone"
  • "Emergency stop button must be red and thumb-accessible"

Requirements Extracted:

  • [SYS-HMI-001] Display confidence level (% not needed, use bar)
  • [SYS-HMI-002] Show battery level graphically
  • [SYS-HMI-003] Emergency stop is red, 3cm diameter, thumb-accessible

4. Observation (Ethnographic Study)

Method: Follow users in their actual work environment

Example: Hospital infusion pump usage

  • Observe nurses preparing and administering infusions
  • Identify pain points (alarms not heard, data entry tedious)
  • Document workarounds (tape over alarms, printed labels)

Requirements Extracted:

  • [SYS-ALARM-001] Alarms must be audible in noisy ICU environment (≥85 dB)
  • [SYS-UI-001] Allow pre-programmed drug library (reduce manual entry)
  • [SYS-LABEL-001] Support barcode scanning for patient/drug verification

Requirements Writing Standards (ISO/IEC 29148)

Good Requirement Characteristics (ISO/IEC 29148:2018):

Attribute Definition Example
Necessary Addresses a stakeholder need [PASS] "System shall detect air bubbles"
Unambiguous Single interpretation [PASS] "Detect 50 µL air within 2 seconds" (NOT "quickly")
Verifiable Can be tested/measured [PASS] "Air detection ≥95% accuracy" (testable)
Bounded Specific scope, no "etc." [PASS] "Support 50 drugs" (NOT "multiple")
Feasible Achievable with known technology [PASS] "±5% accuracy" (feasible)
Complete Not dependent on other unwritten requirements [PASS] All relevant context provided
Consistent No conflicts with other requirements [PASS] "Safety priority" defined once

Requirement Template (ISO/IEC 29148):

[REQ-XXX-NNN] Requirement Title

Description:
  The system/component/interface shall [action]
  [under conditions] to [achieve purpose].

Rationale:
  Why this requirement exists (regulatory, user need, safety)

Verification Method:
  Test / Analysis / Inspection / Demonstration

Derived From:
  Source stakeholder, regulation, or parent requirement

Traceability:
  Forward link: Design elements implementing this
  Backward link: Stakeholder need this addresses

Example Requirement:

[SYS-SAFE-001] Air-in-Line Detection

Description:
  The system shall detect air bubbles ≥50 microliters in the IV tubing
  and halt infusion within 2 seconds of detection.

Rationale:
  Large air embolism (>5 mL) is potentially fatal (FDA guidance).
  50 µL target provides early warning margin.

Verification Method:
  Test: Introduce calibrated air bubbles (20, 50, 100 µL)
  Verify: Detection within 2 seconds in 19/20 trials (95%)

Derived From:
  Stakeholder: ICU nurses
  Regulation: FDA Infusion Pump Guidance
  Risk: [RISK-004] Air embolism hazard

Traceability:
  Forward: [DESIGN-AIR-001] Air detection architecture
  Backward: [RISK-004] Air embolism mitigation

Traceability Baseline

Requirements Traceability Matrix (RTM) Setup:

Stakeholder Requirement → System Requirement → Design Element → Test Case

Example Path:
[STK-REQ-001] "Nurses need to trust infusion"
           ↓
[SYS-REQ-001] "System shall verify drug correctness via barcode"
           ↓
[DESIGN-BarCode] Barcode scanner module
           ↓
[TEST-001] "Verify barcode reading accuracy (99.9%)"

RTM Tool Setup (Jama, DOORS, SpiraTest):

  • Import all stakeholder requirements
  • Create system requirement child-requirements
  • Link to design elements (traceability forward)
  • Link to test cases (verification)
  • Track coverage (% of stakeholder → verified)

AI Assistance:

  • L1: Suggest RTM links based on keyword matching
  • L1: Auto-generate requirement IDs (REQ-001, REQ-002)
  • HITL: Human reviews and validates links

Requirements Prioritization

When conflicting requirements arise, use structured prioritization:

Priority Category Description
M Must Have Non-negotiable, system fails without it
S Should Have Important but not critical
C Could Have Desirable if resources permit
W Won't Have Explicitly out of scope

AI Assistance: AI can suggest initial prioritization based on:

  • Regulatory keywords (shall, must, required)
  • Safety-related terminology
  • Historical priority patterns from similar projects

Regulatory Requirements Handling

Requirements from regulations require special attention:

Source Examples Handling
ISO 26262 Safety goals, ASIL allocation Mark as derived, trace to standard
UNECE WP.29 Cybersecurity, OTA updates Reference specific regulation clauses
EU Type Approval EMC, environmental Link to homologation requirements

Summary

SYS.1 Requirements Elicitation:

  • Primary mode: Human-driven (L0-L1)
  • AI value: Document analysis, draft generation, consistency checking
  • Human essential: Stakeholder interaction, judgment, agreement
  • Key outputs: Stakeholder requirements specification
  • Traceability: To sources and forward to system requirements