Dependable Software Engineering. Theories, Tools, and Applications - 9th International Symposium, SETTA 2023, Nanjing, China, November 27-29, 2023, Proceedings

Sun Jun, Jun Sun, Holger Hermanns, Lei Bu, Bu, LeiSetta 2024 · DOI · DBLP

The 9th International Symposium on Dependable Software Engineering: Theories, Tools, and Applications (SETTA 2023), recently held in Nanjing, China, serves as a critical forum for researchers bridging the divide between mathematically rigorous formal methods and the practical demands of large-scale software engineering. This volume collects 24 high-quality papers selected from 89 submissions, representing a competitive 27% acceptance rate. The proceedings highlight a significant evolution in the field, moving beyond traditional verification toward securing modern, complex infrastructures such as neural networks, distributed coordination services, and autonomous cyber-physical systems.

Addressing the "Dependability Gap"

The central challenge addressed by the SETTA symposium series is the "dependability gap." While formal methods provide the mathematical certainty required for safety-critical systems, they often struggle with the scale and complexity of real-world software. Conversely, standard software engineering practices scale well but often lack the rigor necessary for high-stakes environments. The research presented in this volume seeks to reconcile these two worlds by applying formal logic, model checking, and automated reasoning to contemporary software challenges.

Key Research Frontiers

The symposium’s contributions are categorized into several high-impact areas:

1. AI Safety and Robustness

As deep learning (DL) systems are increasingly deployed in safety-critical roles, verifying their robustness is paramount. Papers in this volume explore "Diversity-Guided" test generation for neural networks and coverage criteria—such as the HeatC method—to ensure DL systems are resilient against adversarial attacks. This convergence of formal methods and AI also includes using adversarial training to improve formal tasks like log anomaly detection.

2. Distributed Systems and Complexity

Ensuring reliability in distributed services remains a core challenge. Researchers utilized TLA+ to specify and verify the ZooKeeper coordination service, identifying potential reliability improvements. Other work focused on the formalization of eBPF virtual machines for microcontrollers and the safety of autonomous spacecraft rendezvous scenarios involving obstacle avoidance.

3. Security and Vulnerability Analysis

Security research in the volume emphasizes behavioral semantics and taint analysis. Notable contributions include concolic execution frameworks for Windows binaries and automated detection of vulnerabilities in web applications. The proceedings also address the security of regular expressions used "in the wild," highlighting the practical utility of formal string analysis.

Methodological Innovations

The papers detail a diverse array of methodologies that blend classical logic with modern computational techniques:

  • Automated Reasoning: Advances in SMT (Satisfiability Modulo Theories) solvers for non-linear real arithmetic and complex string constraints show significant efficiency gains over previous state-of-the-art tools.
  • Formal Modeling: The use of Session Types for communication protocols and Lambda Calculus for reasoning frameworks continues to provide a solid foundation for verifying interaction patterns.
  • Synthesis: One highlight includes the automatic generation of Behavior Trees through formal verification-based synthesis, streamlining the development of autonomous agents.

Results and Impact

The quantitative results reported across the papers underscore the field's progress. New algorithms for solving string constraints have demonstrated higher throughput and lower latency. In the realm of anomaly detection, AI-integrated formal techniques have shown improved accuracy in identifying system log irregularities compared to traditional baseline tools. These results suggest that the integration of formal methods into the software development lifecycle is becoming increasingly viable for industrial applications.

Future Directions

Despite these successes, several challenges persist that define the roadmap for future research:

  • Scalability: The "state space explosion" remains a primary hurdle for applying formal proofs to massive, multi-million-line codebases.
  • Automation: Reducing the manual effort required for writing formal specifications (like TLA+) is essential for broader industry adoption.
  • Dynamic Assurance: As systems move toward reinforcement learning and autonomous adaptation, the community is shifting focus from design-time verification to "run-time" assurance to handle environments that change over time.

SETTA 2023 demonstrates that dependable software engineering is no longer a niche academic pursuit but a necessary discipline for the safe operation of our increasingly digital and autonomous world.

Summary updated

BibTeX
@inproceedings{jun2024dependable,
  author = {Sun Jun and Jun Sun and Holger Hermanns and Lei Bu and Bu, Lei},
  title = {{Dependable Software Engineering. Theories, Tools, and Applications - 9th International Symposium, SETTA 2023, Nanjing, China, November 27-29, 2023, Proceedings}},
  booktitle = {Setta},
  year = {2024},
  doi = {10.1007/978-981-99-8664-4},
}

Loading related content...

Figure