National Key R&D Program Software Testing: Standards, Methods, and Third-Party Acceptance

National Key R&D Program software testing provides independent, standards-based verification of software deliverables developed under government-funded science and technology projects in China. Accredited third-party laboratories conduct functional, performance, security, and reliability testing in accordance with GB/T and ISO/IEC standards, issuing CNAS- and CMA-stamped reports that directly support project acceptance, achievement appraisal, and milestone review.

Technical Overview

China's National Key R&D Programs constitute the country's principal funding channel for major scientific and technological research, consolidating earlier national programs under a unified management framework administered by the Ministry of Science and Technology. Because these programs are supported by public funding, every software deliverable must undergo rigorous verification before a project can pass final acceptance review. National Key R&D Program software testing therefore refers to the independent evaluation of software systems, platforms, and tools developed under these programs, performed by accredited third-party laboratories to confirm that deliverables meet the technical indicators defined in the project task contract.

The core objective of this testing is to provide objective, reproducible evidence of software quality. Testing laboratories evaluate whether the delivered software functions correctly, performs within specified limits, remains secure and reliable, and satisfies the quantitative and qualitative targets committed to by the research team. The evaluation framework is grounded in internationally recognized quality models, principally the ISO/IEC 25010 product quality model and its Chinese national standard adoption, GB/T 25000.51, which defines the requirements and evaluation methods for ready-to-use software products and user documentation.

Only laboratories holding CNAS accreditation in accordance with ISO/IEC 17025, as well as CMA measurement accreditation, are qualified to issue test reports recognized during program acceptance. This accreditation requirement guarantees metrological traceability, technical competence, and impartiality, and ensures that reports issued in China are accepted under international mutual recognition arrangements.

Testing Samples

Software deliverables submitted for National Key R&D Program testing span a wide range of technologies. Typical sample categories include the following.

Application and Business Management Software

Information systems developed for government administration, public services, agriculture, environmental monitoring, transportation, and healthcare. These systems are usually tested for functional completeness, business process correctness, and multi-user concurrent performance.

Embedded and Industrial Control Software

Software embedded in intelligent equipment, instrumentation, robotics, energy systems, and manufacturing lines. Testing focuses on real-time behavior, hardware-software interface correctness, fault tolerance, and stability under continuous operation.

Scientific Computing and Big Data Software

Data processing pipelines, simulation platforms, analytical toolsets, and high-performance computing applications. Evaluation covers computational accuracy against reference datasets, throughput, scalability, and data integrity across large volumes.

Artificial Intelligence and Algorithmic Software

Machine learning frameworks, computer vision modules, natural language processing engines, and intelligent decision-support systems. Testing verifies algorithm performance metrics such as precision, recall, and inference latency, together with functional robustness and, where applicable, conformance to AI-related GB/T standards.

Basic Software and Platform Products

Operating systems, databases, middleware, development environments, and cloud platform components. These products are examined for standards compliance, interface consistency, migration capability, and co-existence with mainstream ecosystem software.

Mobile and Web Applications

Mobile clients, browser-based portals, and hybrid applications delivered as user-facing outputs of research projects, tested across device models, operating system versions, browsers, and network conditions.

Testing Items

Testing items are selected based on the quality characteristics defined in ISO/IEC 25010 and GB/T 25000.51, combined with the specific indicators stated in the project task contract. Common testing items include the following.

Functional Suitability Testing

Verification of functional completeness, functional correctness, and functional appropriateness. Every documented function is traced to requirements and executed against designed test cases, covering normal flows, boundary conditions, and exception handling.

Performance Efficiency Testing

Measurement of time behavior, resource utilization, and capacity under representative and peak workloads. Typical metrics include response time, throughput, transaction success rate, and concurrent user capacity, validated through load, stress, and stability scenarios.

Security Testing

Assessment of confidentiality, integrity, authenticity, accountability, and non-repudiation. Activities include vulnerability scanning, weakness analysis against common vulnerability classifications, verification of authentication and authorization mechanisms, data protection checks, and, where required, evaluation aligned with cybersecurity protection standards such as GB/T 22239 and the GB/T 28448 series.

Reliability and Compatibility Testing

Evaluation of maturity, availability, fault tolerance, and recoverability, including recovery behavior after abnormal termination. Compatibility testing confirms co-existence with other software and interoperability across specified platforms, interfaces, and data exchange protocols.

Usability, Maintainability, and Portability Testing

Inspection of user interface appropriateness, learnability, and operability; analysis of modularity, reusability, and analysability of the code structure; and verification of adaptability, installability, and replaceability across target environments.

Source Code and Documentation Evaluation

Static analysis of source code to detect defects, security weaknesses, excessive complexity, and violations of coding conventions, together with review of user documentation for completeness, correctness, and consistency with actual system behavior, as required by GB/T 25000.51.

Testing Methods

Accredited third-party laboratories apply a combination of dynamic and static methods, structured according to GB/T 15532 for software testing and documented in line with GB/T 9386 and the ISO/IEC/IEEE 29119 series.

• Black-box functional testing: requirement-based test case design using equivalence partitioning, boundary value analysis, decision tables, and scenario-driven methods.
• Automated regression testing: script-based execution to confirm defect fixes and prevent functional regressions across test cycles.
• Performance and load testing: workload simulation with industry-standard tooling to model concurrent users, sustained throughput, and stress conditions.
• Security testing: static and dynamic analysis, configuration review, penetration-style verification, and checks against known vulnerability patterns.
• Static code analysis: automated scanning for code defects, security weaknesses, and maintainability metrics such as cyclomatic complexity and duplication rate.
• Documentation review: systematic comparison of user manuals, installation guides, and design documents against the delivered software.

The standard testing workflow proceeds through the following stages: commissioning and contract confirmation; review of submitted materials including the task contract, requirement specifications, and user documentation; test planning and test case design based on contractual indicators; construction of the test environment; test execution with full evidence capture; defect recording, developer rectification, and regression verification; and finally issuance of the test report bearing CNAS and CMA accreditation marks.

Applications

Test reports generated through this process serve multiple critical purposes within the national science and technology management system.

• Project and subject acceptance: independent test evidence is a mandatory attachment for final acceptance review of projects funded under national key programs.
• Mid-term and milestone review: interim testing verifies progress against phased technical indicators and informs continued funding decisions.
• Scientific and technological achievement appraisal: objective quality data supports expert evaluation of the novelty, maturity, and practical value of research outcomes.
• Awards and follow-on applications: test reports substantiate nominations for science and technology awards and support applications for patents and software copyright registration.
• Productization and market entry: verified quality results accelerate the transition of research prototypes into deployable commercial products and strengthen bidding qualifications.

FAQ

Why is third-party software testing mandatory for National Key R&D Program acceptance?

Program management regulations require that software deliverables be evaluated by an impartial laboratory that holds CNAS and CMA accreditation. This requirement ensures that acceptance decisions rest on objective, technically competent evidence rather than self-declaration by the development team, protecting the integrity of publicly funded research.

Which standards govern the testing process?

The primary standard is GB/T 25000.51, which adopts the SQuaRE requirements for software product quality and documentation. Testing is also conducted with reference to GB/T 15532 for testing specifications, GB/T 9386 for test documentation, ISO/IEC 25010 for the quality model, ISO/IEC/IEEE 29119 for test processes, and, for security-sensitive systems, the GB/T 22239 and GB/T 28448 series.

What materials must a project team prepare before testing?

Typical submissions include the project task contract or statement of work, software requirement specifications, design documents, user and installation manuals, the executable software package or deployment package, database scripts, and a list of contractual technical indicators to be verified. Complete and consistent materials significantly shorten the testing cycle.

How long does the testing cycle take?

For a medium-scale management information system, the cycle typically ranges from three to six weeks, covering test design, execution, defect rectification, and regression verification. Large platforms, embedded systems, or high-performance computing software may require additional time for environment construction and extended stability testing.

Is the test report valid for purposes beyond project acceptance?

Yes. Because the report carries CNAS and CMA marks, it is recognized in achievement appraisal, award applications, software copyright and patent proceedings, procurement qualification, and commercial promotion, provided the tested software version remains unchanged from the evaluated release.