Research Seminar: Jean Baptiste Minani

May 06, 2026

1:00 p.m. ET

A203

IoT systems are increasingly used in many domains, and they require strong end-to-end (E2E) testing to ensure correct functional behavior. However, generating effective E2E tests is difficult due to the heterogeneity, distributed nature, and dynamic execution of IoT systems. It is also challenging to vali-date interactions across multiple layers and components. Traditional testing approaches often fail to expose faults under real-world, cross-layer conditions.

In this talk, I present FUNEETIS, a semi-automated approach for E2E functional testing of IoT systems. It uses Use Case Specifications (UCSs) written in a restricted format and a description of the system under test (SUT). FUNEETIS converts UCSs into executable scenarios, transforms them into structured payloads, and generates executable test cases. To detect faults, it executes the system in real time, collects runtime data across layers, and compares expected and actual results. 

I evaluated FUNEETIS through case studies on two IoT systems. Faults with known locations were injected to establish ground truth, allowing us to measure bug detection precision and recall, along with IoT-specific coverage metrics. The results show full coverage of nodes, protocols, and scenarios, and high accuracy in scenario extraction and test data generation, with precision and recall generally above 90%. Interaction and action coverage are moderate, and most detected bugs are located in the device and application layers across both systems. FUNEETIS relies only on structured UCSs and a sys-tem description, and it can be adapted to other IoT systems.

More recently, I have been exploring the use of generative AI, especially LLM-based agent architectures, to automatically analyze technical documentation, construct a Knowledge Graph, and then use the generated graph to create tests. This work moves beyond restricted inputs and aims to support more general software systems.

As a future direction, I am investigating a hybrid approach that converts unstructured requirements into structured representations and then generates tests using multiple agents. Research Seminar

Bio

Jean Baptiste Minani is a Postdoctoral Researcher at the University of Ottawa, Canada, where his research focuses on generative AI for software engineering, particularly LLM-based agent architectures for automated test generation and software quality assurance. He completed his Ph.D. in Soft-ware Engineering at Concordia University, Canada. He holds a master’s degree in information technology from Carnegie Mellon University (CMU), USA, and a Bachelor’s degree in Information Technology from Vellore Institute of Technology (VIT), India. From 2010 to 2021, he worked in the field of software engineering, specializing in large-scale e-government systems. During this period, he led and contributed to the development of several national digital platforms by applying modern software engineering practices to improve service delivery across government-to-government (G2G), government-to-citizen (G2C), and government-to-business (G2B) interactions. His research interests include automated software testing, end-to-end testing of complex and distributed systems (including IoT systems), and the use of artificial intelligence to improve software quality. More recently, his work explores generative AI and LLM-agent architectures for analyzing technical documentation, extracting structured knowledge, and using the extracted information to generate tests. He has published research in leading journals and conferences in software engineering and IoT systems, contributing to advances in automated testing and quality assurance for modern soft-ware systems.

Upcoming Events