Comparison and Benchmarking of TSNsched vs. YAS

While existing tools like TSNsched (implemented in Java and leveraging Z3 for constraint solving) provide a foundation for TSN scheduling, the development of a new scheduler in Rust with libSMT introduces novel opportunities for performance optimization, scalability, and integration with modern systems. Both schedulers can be invoked through multiple interfaces, but their underlying implementation differences — such as language efficiency, library integration, and computational overhead — significantly impact their suitability for large-scale TSN networks.

The performance of these schedulers depends on factors like constraint solving speed, memory usage, latency in scheduling decisions, and compatibility with real-time requirements. By evaluating these metrics, researchers can determine whether the Rust-based scheduler with libSMT offers advantages over TSNsched in terms of speed, resource efficiency, and flexibility.

Goals of the thesis

The goal of this thesis is to conduct a comprehensive performance evaluation of two TSN schedulers — TSNsched (Java + Z3) and a new Rust-based scheduler (libSMT) — by comparing their speed, resource usage, and real-time capabilities. The project will also investigate the integration and calling mechanisms of both schedulers, assessing how they can be embedded into network architectures or used as standalone tools.

Requirements

• Experience with Rust, perf (and Java)

Contact

• J. Bigge
• C. Sommer