Search Results

Documents authored by Azumi, Takuya

Document
DOI: 10.4230/LIPIcs.ECRTS.2026.3
Probabilistic Schedulability Analysis for Mixed-Criticality DAG Tasks on Multiprocessors

Authors: Hiroto Takahashi, Atsushi Yano, and Takuya Azumi

Published in: LIPIcs, Volume 375, 38th European Conference on Real-Time Systems (ECRTS 2026)

Abstract
Mixed-criticality DAG task systems on multiprocessors require schedulability analysis to guarantee that safety-critical tasks meet their deadlines. Conventional approaches rely on worst-case execution times (WCETs), which account for extremely rare pathological scenarios and consequently lead to significant over-provisioning of processor cores. This paper proposes a probabilistic schedulability analysis that exploits the statistical rarity of multiple vertices within a DAG exceeding their expected execution budgets. By grouping vertices within each DAG into small clusters and bounding the probability that more than one vertex in a cluster overruns, the method assigns each cluster a tighter execution budget than the sum of individual WCETs, thereby reducing the number of cores required. The clustering configuration is optimized via simulated annealing to minimize total core usage while maintaining a designer-specified bound on the system-level probability of deadline misses. Experiments on synthetic task sets demonstrate that the proposed method reduces the required number of cores by up to 36% compared to a deterministic baseline, with larger gains for DAGs exhibiting higher internal parallelism.
Cite as

Hiroto Takahashi, Atsushi Yano, and Takuya Azumi. Probabilistic Schedulability Analysis for Mixed-Criticality DAG Tasks on Multiprocessors. In 38th European Conference on Real-Time Systems (ECRTS 2026). Leibniz International Proceedings in Informatics (LIPIcs), Volume 375, pp. 3:1-3:23, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)

Copy BibTex To Clipboard

@InProceedings{takahashi_et_al:LIPIcs.ECRTS.2026.3,
  author =	{Takahashi, Hiroto and Yano, Atsushi and Azumi, Takuya},
  title =	{{Probabilistic Schedulability Analysis for Mixed-Criticality DAG Tasks on Multiprocessors}},
  booktitle =	{38th European Conference on Real-Time Systems (ECRTS 2026)},
  pages =	{3:1--3:23},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-95977-429-1},
  ISSN =	{1868-8969},
  year =	{2026},
  volume =	{375},
  editor =	{Kritikakou, Angeliki},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.ECRTS.2026.3},
  URN =		{urn:nbn:de:0030-drops-265871},
  doi =		{10.4230/LIPIcs.ECRTS.2026.3},
  annote =	{Keywords: Mixed-Criticality Systems, DAG Tasks, Probabilistic Analysis, Federated Scheduling, Real-Time Systems}
}
Document
Artifact
DOI: 10.4230/DARTS.12.2.11
Probabilistic Schedulability Analysis for Mixed-Criticality DAG Tasks on Multiprocessors (Artifact)

Authors: Hiroto Takahashi, Atsushi Yano, and Takuya Azumi

Published in: DARTS, Volume 12, Issue 2, Special Issue of the 38th European Conference on Real-Time Systems (ECRTS 2026)

Abstract
This artifact accompanies the ECRTS 2026 paper "Probabilistic Schedulability Analysis for Mixed-Criticality DAG Tasks on Multiprocessors". The package provides the implementation of the proposed probabilistic schedulability analysis and baseline analyses, together with the DAG-generation wrapper, experiment drivers, plotting scripts, and input data needed to reproduce the main evaluation results of the paper. The primary reproduction workflow covers the paper-scale experiments for Figures 3-5, while additional check and smoke modes provide faster functional validation. The artifact was evaluated in the ECRTS 2026 Artifact Evaluation process and received all four artifact-evaluation seals: Artifact Available, Artifact Evaluated - Functional, Artifact Evaluated - Reusable, and Artifact Evaluated - Results Reproduced.
Cite as

Hiroto Takahashi, Atsushi Yano, and Takuya Azumi. Probabilistic Schedulability Analysis for Mixed-Criticality DAG Tasks on Multiprocessors (Artifact). In Special Issue of the 38th European Conference on Real-Time Systems (ECRTS 2026). Dagstuhl Artifacts Series (DARTS), Volume 12, Issue 2, pp. 11:1-11:3, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)

Copy BibTex To Clipboard

@Article{takahashi_et_al:DARTS.12.2.11,
  author =	{Takahashi, Hiroto and Yano, Atsushi and Azumi, Takuya},
  title =	{{Probabilistic Schedulability Analysis for Mixed-Criticality DAG Tasks on Multiprocessors (Artifact)}},
  pages =	{11:1--11:3},
  journal =	{Dagstuhl Artifacts Series},
  ISSN =	{2509-8195},
  year =	{2026},
  volume =	{12},
  number =	{2},
  editor =	{Takahashi, Hiroto and Yano, Atsushi and Azumi, Takuya},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/DARTS.12.2.11},
  URN =		{urn:nbn:de:0030-drops-266280},
  doi =		{10.4230/DARTS.12.2.11},
  annote =	{Keywords: Mixed-Criticality Systems, DAG Tasks, Probabilistic Analysis, Federated Scheduling, Real-Time Systems, Artifact Evaluation}
}
Document
DOI: 10.4230/LIPIcs.ECRTS.2024.8
Deadline Miss Early Detection Method for DAG Tasks Considering Variable Execution Time

Authors: Hayate Toba and Takuya Azumi

Published in: LIPIcs, Volume 298, 36th Euromicro Conference on Real-Time Systems (ECRTS 2024)

Abstract
Autonomous driving systems must guarantee safety, which requires strict real-time performance. A series of processes, from sensor data input to vehicle control command output, must be completed by the end-to-end deadline. If a deadline miss occurs, the system must quickly transition to a safe state. To improve safety, an early detection method for deadline misses was proposed. The proposed method represents the autonomous driving system as a directed acyclic graph (DAG) with a mixture of timer-driven and event-driven nodes. It assigns appropriate time constraints for each node based on the end-to-end deadline. However, the existing methods assume the worst-case execution time (WCET) for calculating the time constraints of each node and do not consider the execution time variation of nodes, making the detection of deadline misses pessimistic. This paper proposes a deadline miss early detection method to determine the possibility of deadline misses quantitatively at the beginning of each node execution in a DAG task. It calculates the time constraints of each node using probabilistic execution time, which treats execution time as a random variable. Experimental evaluation shows that the proposed method reduces pessimism, which is a problem of conventional methods using WCET, and then achieves more accurate early detection of deadline misses. The evaluation also indicates that the execution time of static analysis required for deadline miss early detection is within a practical level.
Cite as

Hayate Toba and Takuya Azumi. Deadline Miss Early Detection Method for DAG Tasks Considering Variable Execution Time. In 36th Euromicro Conference on Real-Time Systems (ECRTS 2024). Leibniz International Proceedings in Informatics (LIPIcs), Volume 298, pp. 8:1-8:21, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2024)

Copy BibTex To Clipboard

@InProceedings{toba_et_al:LIPIcs.ECRTS.2024.8,
  author =	{Toba, Hayate and Azumi, Takuya},
  title =	{{Deadline Miss Early Detection Method for DAG Tasks Considering Variable Execution Time}},
  booktitle =	{36th Euromicro Conference on Real-Time Systems (ECRTS 2024)},
  pages =	{8:1--8:21},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-95977-324-9},
  ISSN =	{1868-8969},
  year =	{2024},
  volume =	{298},
  editor =	{Pellizzoni, Rodolfo},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.ECRTS.2024.8},
  URN =		{urn:nbn:de:0030-drops-203116},
  doi =		{10.4230/LIPIcs.ECRTS.2024.8},
  annote =	{Keywords: Autonomous driving system, deadline miss early detection, DAG, event-driven task, timer-driven task, probabilistic execution time}
}
Document
DOI: 10.4230/OASIcs.ASD.2019.3
IDF-Autoware: Integrated Development Framework for ROS-Based Self-Driving Systems Using MATLAB/Simulink

Authors: Shota Tokunaga, Yuki Horita, Yasuhiro Oda, and Takuya Azumi

Published in: OASIcs, Volume 68, Workshop on Autonomous Systems Design (ASD 2019)

Abstract
This paper proposes an integrated development framework that enables co-simulation and operation of a Robot Operating System (ROS)-based self-driving system using MATLAB/Simulink (IDF-Autoware). The management of self-driving systems is becoming more complex as the development of self-driving technology progresses. One approach to the development of self-driving systems is the use of ROS; however, the system used in the automotive industry is typically designed using MATLAB/Simulink, which can simulate and evaluate the models used for self-driving. These models are incompatible with ROS-based systems. To allow the two to be used in tandem, it is necessary to rewrite the C++ code and incorporate them into the ROS-based system, which makes development inefficient. Therefore, the proposed framework allows models created using MATLAB/Simulink to be used in a ROS-based self-driving system, thereby improving development efficiency. Furthermore, our evaluations of the proposed framework demonstrated its practical potential.
Cite as

Shota Tokunaga, Yuki Horita, Yasuhiro Oda, and Takuya Azumi. IDF-Autoware: Integrated Development Framework for ROS-Based Self-Driving Systems Using MATLAB/Simulink. In Workshop on Autonomous Systems Design (ASD 2019). Open Access Series in Informatics (OASIcs), Volume 68, pp. 3:1-3:9, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2019)

Copy BibTex To Clipboard

@InProceedings{tokunaga_et_al:OASIcs.ASD.2019.3,
  author =	{Tokunaga, Shota and Horita, Yuki and Oda, Yasuhiro and Azumi, Takuya},
  title =	{{IDF-Autoware: Integrated Development Framework for ROS-Based Self-Driving Systems Using MATLAB/Simulink}},
  booktitle =	{Workshop on Autonomous Systems Design (ASD 2019)},
  pages =	{3:1--3:9},
  series =	{Open Access Series in Informatics (OASIcs)},
  ISBN =	{978-3-95977-102-3},
  ISSN =	{2190-6807},
  year =	{2019},
  volume =	{68},
  editor =	{Saidi, Selma and Ernst, Rolf and Ziegenbein, Dirk},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.ASD.2019.3},
  URN =		{urn:nbn:de:0030-drops-103367},
  doi =		{10.4230/OASIcs.ASD.2019.3},
  annote =	{Keywords: self-driving systems, framework, robot operating system (ROS), MATLAB/Simulink}
}
Any Issues?
X

Feedback on the Current Page

CAPTCHA

Thanks for your feedback!

Feedback submitted to Dagstuhl Publishing

Could not send message

Please try again later or send an E-mail
© 2023-2026 Schloss Dagstuhl – LZI GmbH Schloss Dagstuhl – LZI GmbH About DROPS Imprint Privacy Contact