Model-Based Testing of Zero-Copy Protocols

From ISLAB/CAISR
Title Model-Based Testing of Zero-Copy Protocols
Summary Challenges in Model-Based Testing of Zero-Copy Protocols
Keywords model-based testing, zero-copy protocols, bulk-data communication, test coverage
TimeFrame 6 months, starting as soon as possible
References 1. Jan Tretmans. Model-based testing and some steps towards test-based modelling. In Marco Bernardo and Valérie Issarny, editors, Formal Methods for Eternal Networked Software Systems, LNCS 6659, pages 297–326. Springer, 2011.

2. Wojciech Mostowski, Thomas Arts, and John Hughes. Modelling of Autosar Libraries for Large Scale Testing. Proceedings, 2nd Workshop on Models for Formal Analysis of Real Systems (MARS), EPTCS 244, 2017.

3. ICEORYX home page: https://iceoryx.io/

4. ALEX.AI home page: https://www.apex.ai/

Prerequisites Good programming skills, preferably the course on Testing and Verification of Embedded Systems complete, or similar.
Author
Supervisor Wojciech Mostowski
Level Master
Status Open


Zero-Copy protocol is an inter-process communication paradigm where the transmitted data is not replicated for transmission, but rather provided "referentialy". This is done to improve performance, which is necessary for certain kinds of applications, especially in AI, where a lot of data has to be exchanged and communication channels are the major bottleneck. Zero-Copy protocols also bring new challenges, which include memory safety (concurrent access, race-conditions), data expiry / availability, and real-time requirements (time-performance guarantees).

This project shall be done in co-operation with APEX.AI, a major software provider for vehicular applications (including "automotive" ROS), and will target ICEORYX, an open-source implementation / platform for Zero-Copy protocols. The core question of the project is how can specific properties of Zero-Copy protocols be modeled / specified to enable Model-Based Testing (MBT), an automated testing method where tests (input test data and oracles) are generated from the model.

Additional supervisors: Jan Tretmans (TNO, NL), Anders Kallerdahl (APEX.IO, Gothenburg)