The Time-Triggered Wireless Architecture (Artifact)

Romain Jacob, Licong Zhang, Marco Zimmerling, Jan Beutel, Samarjit Chakraborty, Lothar ThieleDagstuhl Artifacts Ser 2020 · OpenReview

TL;DR

This paper presents a reproducible artifact for the Time-Triggered Wireless Architecture (TTW), a system designed for reliable and synchronized communication in wireless sensor networks. The artifact provides all necessary code and data to replicate experiments, enabling researchers to validate and build upon the TTW system.

What Problem Does This Paper Address?

Developing reliable real-time systems, especially in wireless environments, is a significant challenge. Traditional wireless communication often suffers from unpredictable delays and interference, making it unsuitable for applications requiring strict timing guarantees (e.g., industrial control, automotive systems). The Time-Triggered Wireless Architecture (TTW) aims to address this by introducing a deterministic communication mechanism, similar to wired time-triggered systems, but adapted for the wireless domain. This artifact focuses on making the TTW system and its evaluation reproducible, a crucial step for scientific validation and community adoption.

Methods

The artifact provides implementations for several key components of the TTW system:

  • TTnet: This is the core wireless network stack that implements the time-triggered communication protocol. The artifact includes its C source code, with instructions for compilation and remote execution on the FlockLab testbed.
  • TTnet Model: A time and energy model for TTnet is provided as Python code. This model can be run directly in a web browser using Binder or locally by installing dependencies listed in a requirements.txt file.
  • TTW Scheduler: This component is responsible for generating the communication schedules. It's implemented in Matlab and relies on the Gurobi solver for its Mixed Integer Linear Programming (MILP) formulation. This is noted as the most challenging part of the artifact to re-run due to its reliance on commercial software.
  • Data Processing and Plotting Scripts: Python scripts are included for processing raw data from TTnet model validation and for generating the plots presented in the original article. These are also accessible via Binder or local installation.

Datasets

The artifact includes:

  • Raw data from FlockLab experiments: Collected over several months, this data is provided to allow for eventual reproduction of networking experiments.
  • FlockLab configuration files: These files were used to obtain the raw data and are included to aid in reproduction.

The artifact itself is not a new benchmark dataset but rather a collection of code and data designed to reproduce experiments related to the TTW system.

Key Insights and Ideas

  • Bridging the Gap to Determinism: The core idea is to bring the predictability of wired time-triggered systems to the inherently less predictable wireless domain. This is achieved through a carefully designed protocol and scheduling mechanism.
  • Reproducibility as a Cornerstone: The emphasis on providing a stable, versioned artifact with clear instructions highlights the growing importance of reproducibility in scientific research, especially for complex systems.
  • Modular Design: The artifact is broken down into distinct components (network stack, models, scheduler, data processing), allowing researchers to potentially focus on and reuse specific parts.
  • Leveraging Existing Infrastructure: The use of FlockLab for networking experiments and Binder for code execution demonstrates a practical approach to making complex systems accessible.

Key Results

The artifact's primary contribution is not a set of new quantitative results but rather the ability to reproduce the results of the original article. This includes:

  • Replication of computational experiments.
  • Reproduction of TTW schedule derivations.
  • Validation of the TTnet model.
  • Generation of all plots presented in the article.

The artifact is available under the GPL-3.0 license and has a size of 7.6 MB.

Limitations and Future Work

  • Commercial Software Dependency: The TTW scheduler's reliance on Matlab and Gurobi makes it the most difficult component to re-run, potentially limiting its accessibility for researchers without licenses.
  • FlockLab Experiment Replication: While raw data and configuration files are provided, replicating the networking experiments on FlockLab itself might still be challenging due to the nature of testbed availability and the complexity of the experiments run over "several months."
  • Future Work (Implied): The availability of this artifact suggests that future work could involve extending the TTW architecture, improving its efficiency, or applying it to new domains. The modular nature of the artifact also facilitates integration with other research efforts.

Summary updated

Abstract

This artifact contains a stable version of all the data and source code required to reproduce or replicate the results presented in The Time-Triggered Wireless Architecture. One GitHub repository serves as main hub for all information related to the artifact. The README file contains detailed instructions for - Running the TTnet model - Compiling and running TTnet - Running the TTW scheduler - Reproducing the data processing - Reproducing the plots
BibTeX
@inproceedings{jacob2020time,
  author = {Romain Jacob and Licong Zhang and Marco Zimmerling and Jan Beutel and Samarjit Chakraborty and Lothar Thiele},
  title = {{The Time-Triggered Wireless Architecture (Artifact)}},
  booktitle = {Dagstuhl Artifacts Ser},
  year = {2020},
  url = {https://openreview.net/forum?id=cXdFnSMun1w},
}
Figure