Assessing The Quality of Smart Ecosystems

Professor Mauro Pezzè presented a comprehensive research agenda focused on software testing and quality assurance in smart, human-centric ecosystems. Framing smart cities like Zurich and Singapore as real-world examples, he emphasized that such ecosystems are not traditional software systems—they involve unpredictable interactions between autonomous systems, human agents, and dynamically evolving environments. 

The challenge lies in testing these systems without fully defined requirements and under emergent, often contradictory behaviors. His team’s solution: move from traditional correctness-based software validation to a "system health" approach—characterizing systems not by perfection, but by their resilience and adaptive capabilities. The research introduced several pillars of this research initiative: autonomic testing using generative AI to continuously update test suites; smart monitoring of complex systems through machine learning-based anomaly detection; and digital mirrors, advanced simulators that model both technological and human behavior. 

Professor Pezzè’s team also showcased how they model social group behavior, applying psychological profiling and behavioral data to forecast human interactions in mobility ecosystems. From real-time failure prediction in microservices to simulating cascading disruptions in urban systems, the research proposes a new paradigm for validating large-scale, decentralized, and adaptive ecosystems—where ensuring "healthy behavior" is more feasible and meaningful than seeking deterministic correctness.