580 California St., Suite 400
San Francisco, CA, 94104
Key research themes
1. How can automated and semi-automated approaches improve consistency management in industrial continuous model-based development?
This theme addresses the practical challenges and desired advancements in achieving consistency among models and other development artifacts within industrial continuous model-based development (MBD) settings. It is crucial because consistency management is a major impediment to adopting more iterative and agile modeling practices that aim to shorten development cycles. Understanding current industrial practices and how automation and migration pathways can facilitate more effective consistency management can inform tool support and methodological improvements.
Key finding: The study of thirteen industrial MBD settings reveals that immature modeling practices hinder the adoption of shorter development cycles, with consistency management being a major challenge. The findings identify that... Read more
Key finding: The proposed XRTSDIC framework introduces an extensible and flexible approach for real-time detection of software design inconsistencies that supports multiple modeling tools, rule languages, and visualization techniques. Its... Read more
Key finding: By utilizing dynamic logic to define a uniform semantic framework for UML diagrams, including class, state machine, and sequence diagrams, this work enables precise formal reasoning about model consistency. The approach... Read more
2. What methodologies and algorithms mitigate state space explosion in model checking for eventual properties through divide-and-conquer and parallelization?
Model checking eventual properties often suffers from state space explosion, limiting practical verification of complex systems. This theme investigates divide-and-conquer based methods that partition the model checking problem into smaller, manageable subproblems, enabling more efficient analysis. The exploration of designing sequential and parallel tools based on these partitions furthers the scalability and running performance of model checking, especially leveraging multi-core architectures for eventual properties represented in temporal logic.
Key finding: The presented divide-and-conquer technique partitions the reachable state space into layered subspaces, enabling equivalence between checking the original eventual property and solving multiple smaller model checking... Read more
Key finding: Building on the divide-and-conquer approach, this work develops a parallel model checking tool leveraging a master-worker pattern on Maude, significantly improving running performance for eventual properties. Experimental... Read more
Key finding: For the related class of conditional stable properties relevant to self-stabilizing systems, this research presents both sequential and parallel tools implementing divide-and-conquer model checking. It demonstrates that the... Read more
3. How can application-centric consistency models in geo-replicated storage systems guarantee invariant preservation with low latency under eventual consistency?
This theme explores consistency models that extend eventual consistency by explicitly considering application-specific invariants to improve correctness guarantees in geo-replicated distributed storage while maintaining low latency. The focus lies on methodologies enabling safe operation execution under weak coordination or repair-based invariant restoration. Middleware implementations demonstrate practical trade-offs in latency and correctness, bridging the gap between strong consistency and eventual consistency for large-scale cloud applications.
Key finding: The paper proposes Explicit Consistency, a consistency model defined by application invariants rather than execution order, allowing operations to execute locally under causal consistency while preserving invariants via a... Read more
Key finding: Confirming the feasibility of Explicit Consistency, this work details static analysis identifying unsafe operations and methods for invariant violation avoidance or repair. It proves that the approach permits local operation... Read more
Key finding: This study extends causally consistent models with temporal constraints, proposing Timed Causal Consistency to achieve convergence—a key property ensuring replicas eventually stabilize to the same state. It analyzes how... Read more