Model-Based Structured Requirements (MBSR)

Updated 1 March 2026

Model-Based Structured Requirements (MBSR) are methodologies and languages for encoding requirements in formal, machine-interpretable system models, ensuring consistency and traceability.
MBSR frameworks leverage INCOSE and ISO/IEC/IEEE standards with SysML extensions to automate completeness checks, traceability, and the generation of test models.
Applications in aerospace, automotive, and MBSE workflows demonstrate significant gains in clarity, reduced rework, and improved verification and validation processes.

Model-Based Structured Requirements (MBSR) constitute a class of methodologies, languages, and tool-supported profiles that encode requirements in formally structured, machine-interpretable forms directly within a system model. MBSR approaches have gained prominence due to their ability to unify natural-language specifications, enforce syntactic and semantic patterns (notably those of INCOSE GtWR and ISO/IEC/IEEE 29148), support requirements verification and validation (V&V), automate traceability, and streamline generation of test models and simulation-ready scenarios. Representative frameworks and empirical data come from domains such as complex aerospace systems (notably NASA JPL), automotive distributed systems, and pan-European model-based engineering projects (Wheaton et al., 2024, Wheaton et al., 2024, Ji et al., 2023, Wiecher et al., 2022, Fischbach et al., 2019, Berger et al., 2019, Sadovykh et al., 2021).

1. Formalization and Structuring of Requirements

At the core of MBSR is the movement from unstructured textual requirements towards formally schematized and ontology-backed artifacts within MBSE tools. The INCOSE-derived MBSR SysML profile, for instance, extends the SysML Requirement metaclass via a set of stereotypes and property tags encapsulating the 42 “Rules”, 49 “Attributes”, and 15 “Characteristics” mandated by the INCOSE Guide to Writing Requirements (GtWR) and mapped to ISO/IEC/IEEE 29148 statement templates (Wheaton et al., 2024, Wheaton et al., 2024). The Profile includes consecutive stereotype layers:

«Requirement_Expression»: Represents a requirement with unique Id, structured natural language “Text”, and five ISO pattern slots (SR1-Condition, SR2-Subject, SR3-Action, SR4-Object, SR5-Constraint_of_Action).
«Requirement_Set»: Allows hierarchical grouping.
«Rule» and «Characteristic»: Explicit model elements used for V&V assertions and compliance tracking.
Complete attribute sets for rationale, traceability, V&V allocation, maturity, priority, status, etc., all to be filled directly in the SysML environment.

Each requirement must populate all prescribed slots (SR1–SR5) according to the enforced pattern, and completeness can be assessed via tool metrics. The default pattern templates enforced at the profile and tool level are:

Default functional:

$\underbrace{\mathrm{[Subject]}_{\mathrm{SR2}}\,\text{shall}\,\underbrace{\mathrm{[Action]}_{\mathrm{SR3}}\,\underbrace{\mathrm{[Constraint]}_{\mathrm{SR5}}}}$

Condition-based:

$\underbrace{\mathrm{[Condition]}_{\mathrm{SR1}}},\,\underbrace{\mathrm{[Subject]}_{\mathrm{SR2}}\,\text{shall}\,\underbrace{\mathrm{[Action]}_{\mathrm{SR3}}\,\underbrace{\mathrm{[Object]}_{\mathrm{SR4}}\,\underbrace{\mathrm{[Constraint]}_{\mathrm{SR5}}}}$

Attributes and rule satisfaction are tracked by formal «satisfy» and «violate» dependencies, and completeness matrices and custom diagrams allow identification of non-conforming requirements and V&V gaps (Wheaton et al., 2024).

2. MBSR in Model-Centric Engineering Workflows

The practical application of MBSR involves importing initial textual or semi-structured requirements, populating meta-model slots and attribute tags, and linking requirements directly to model elements (e.g., SysML Blocks), V&V artifacts (e.g., TestCases), and organizational glossaries (via Allocate relationships) (Wheaton et al., 2024). In the NASA JPL case studies, the adoption of the INCOSE-derived Profile yielded >90% completeness in pattern slot coverage, 100% traceability to system blocks, reduction in ambiguity, and a measurable decrease in downstream rework (Wheaton et al., 2024).

In automotive MBD, pattern-based MBSR languages—restricted fragments of metric temporal logic with unambiguous syntax—can be directly translated into the specification languages of multiple analysis tools. Requirements specified in a pattern language 𝓛 are mapped bijectively to proof objectives or test stubs in Simulink Design Verifier, BTC EmbeddedValidator, and model checkers, thereby supporting seamless, bidirectional formal verification and test generation without redundant manual translations (Berger et al., 2019).

Scenario-based MBSE flows further intertwine requirements, test cases, and system models: requirements are iteratively developed and refined in lock-step with executable test scenarios (using scenario modeling languages such as SMLK), and every requirement is guaranteed bidirectional traceability to both stakeholder concerns and concrete test environments (Wiecher et al., 2022).

3. Algorithms and Automation in Structured Requirement Extraction

A significant contribution of MBSR is the partial automation of requirements structuring from semi-structured artifacts. Automated pipelines decompose raw requirement documents into lines, classify lines as “pseudo-code” (semi-structured) using trained Random Forest classifiers over real-valued feature vectors (special character count, word count, indentation depth, capitalization ratio), and then apply rule-based ANTLR grammars to translate business logic into cause-effect graphs (CEGs) (Fischbach et al., 2019). The formal triple for a CEG is $G=(C,E,R)$ , where $C$ is the set of causes, $E$ the set of effects, and $R$ the directed relations. The approach leads to time savings near 86% when compared with manual model construction, while maintaining structural equivalence between manual and tool-generated CEGs.

A canonical pseudocode fragment for detection is:

def DetectPseudoCode(lines):
    features = []
    for line in lines:
        s = countSpecialChars(line)
        w = countWords(line)
        d = countLeadingSpaces(line)
        r = countCaps(line) / countLetters(line)
        features.append([s, w, d, r])
    labels = RFClassifier.predict(features)
    return [line for (line, lab) in zip(lines, labels) if lab == 1]

with classifier precision and recall metrics carefully optimized for high recall to prevent omission of translatable requirements (Fischbach et al., 2019).

4. Mathematical Foundations for Consistency, Synthesis, and Conflict Resolution

MBSR approaches now encompass rigorous, model-theoretic methods for synthesizing consistent system requirements from multiple siloed stakeholder models. Each stakeholder’s requirement model is encoded as a directed graph (adjacency matrix $M_i$ ), and all pairwise logical dependencies (equivalence, hierarchy, implication, contradiction) between models are encoded as incidence matrices $D_{i,j}$ . Synchronization is formalized: a collection is rationalized when only equivalence and null relations remain in all $D_{i,j}$ , and then synthesis merges equivalent entities to yield a unified and fully traceable requirements view (Ji et al., 2023). This model ensures:

Rigorous conflict detection ( $d^{(i,j)}_{xy}=C$ )
Minimal engineer intervention (synchronization focuses analyst action where logic mismatches exist)
Automated traceability by row/column index preservation
Iterative support for stakeholder change-impact management and re-synchronization as models evolve.

5. Tooling, Automation, and Case Studies

MBSR’s efficacy relies on sophisticated tooling that integrates with leading MBSE environments. For INCOSE-derived patterns, the Model-Based Structured Requirements Profile is delivered as a Cameo Shared Profile for No Magic Cameo Systems Modeler, with:

Tagged values, pattern slots, and rules encoded via stereotype extensions and automatic completeness checks
Custom diagrams (Relation-Map, Dependency Matrix) for V&V, status, and impact analysis
Scripting support (Groovy, OCL) for batch metric calculation, attribute updates, and automated quality checks
Glossary integration and hyper-linkage with underlying architecture
Reporting templates for Word, Excel, and Web Reports.

Case studies from NASA JPL (Mars Sample Return, Thrust Reverser System) and Ford/automotive suppliers demonstrate significant improvements in coverage, clarity, review cycle efficiency, and downstream defect reduction (Wheaton et al., 2024, Berger et al., 2019, Wiecher et al., 2022). In large collaborative European projects, scalable MBSR/MBRE metamodels and custom document-generation pipelines have supported hundreds of contributors and automatic production of complex SRS deliverables (Sadovykh et al., 2021).

Tooling deficits persist in refresh performance for large matrices, need for custom scripting, and complexity of initial attribute population, prompting further development in SysML v2 profiles and digital thread/toolchain integration (Wheaton et al., 2024).

6. Limitations and Future Directions

Despite quantifiable benefits, several challenges persist: performance bottlenecks in large models, manual management of versioning and change-control attributes, limited ReqIF/XMI export fidelity, and the learning curve for full attribute compliance. Migrating MBSR profiles to SysML v2, developing server-side validation and metric computation, refining natural language slot fillers, and integrating with external ALM/version control systems are identified as strategic next steps (Wheaton et al., 2024).

A plausible implication is that, as next-generation MBSE tools mature and standard APIs become ubiquitous, the strict enforcement and live automation of requirement structuring, quality checking, and traceability offered by MBSR will further close the gap between requirements engineering and design verification, offering a digital thread from stakeholder intent to formally verified system build.

Key References

"Digital requirements engineering with an INCOSE-derived SysML meta-model" (Wheaton et al., 2024, Wheaton et al., 2024)
"Automated Generation of Test Models from Semi-Structured Requirements" (Fischbach et al., 2019)
"Multiple Analyses, Requirements Once" (Berger et al., 2019)
"Requirements Rationalization and Synthesis enabled by Model Synchronization" (Ji et al., 2023)
"Model-based Analysis and Specification of Functional Requirements and Tests for Complex Automotive Systems" (Wiecher et al., 2022)
"Applying Model-based Requirements Engineering in Three Large European Collaborative Projects" (Sadovykh et al., 2021)