An Architecture to Enable Machine-Learning-Based Task Migration for Multi-Core Real-Time Systems

Delgadillo, Octavio; Blieninger, Bernhard; Kuhn, Juri; Baumgarten, Uwe

doi:10.1109/mcsoc51149.2021.00066

Cited by 1 publication

(5 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, our publication in MCSoC 2021 [10] introduced the general architecture of the ECU system used in this setup. However, it did not provide a deeper understanding of the machine learning component and the deployment toolchain of the setup, nor did it compare the SNN approach to other approaches, such as the LSTM presented here.…”

Section: Related Workmentioning

confidence: 99%

“…The ECUs are responsible for executing the real-time tasks assigned to them by the task deployment toolchain. The details on their architecture and the tasks executed were presented in the original publication [10]. Therefore, here, we only provide a summary as a context for the rest of the work presented.…”

Section: Ecu System Implementationmentioning

confidence: 99%

“…In order to compare and evaluate the results of our previous work [10], we designed an LSTM network that takes the task parameters for predicting the feasibility of a task set. The features extracted from the task parameters and fed to the LSTM model as input are the priority of the task, its utilization, its relative deadline, and the number of jobs to be executed for a predefined maximum of eight tasks per task set.…”

Section: Lstm Networkmentioning

confidence: 99%

“…These data were generated by deploying random sets of tasks to the ECUs, letting them run for a time span of 60 to 120 s, and then, collecting the data reported by the ECU system on the deadlines missed and the number of finished jobs. As each task set represents one sample, the generation of samples is slow, and therefore, at the time of developing the networks, the task set was relatively small (6000 samples), as mentioned in the original publication [10]. Further data generation allowed us to collect a total of around 29,900 samples.…”

Section: Machine Learningmentioning

confidence: 99%

“…The work in this paper is based on our previous work published in MCSoC 2021 [10] and extends it by describing the setup used to test the architecture introduced there, especially by describing the task deployment toolchain (Section 4) and the machine learning approaches used, which are the spiking neural network (SNN), briefly mentioned in the original publication, and a long-short-term-memory (LSTM) network, introduced here (Section 5).…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

A Generalistic Approach to Machine-Learning-Supported Task Migration on Real-Time Systems

Delgadillo

Blieninger

Kuhn

et al. 2022

JLPEA

Self Cite

View full text Add to dashboard Cite

Consolidating tasks to a smaller number of electronic control units (ECUs) is an important strategy for optimizing costs and resources in the automotive industry. In our research, we aim to enable ECU consolidation by migrating tasks at runtime between different ECUs, which adds redundancy and fail-safety capabilities to the system. In this paper, we present a setup with a generalistic and modular architecture that allows for integrating and testing different ECU architectures and machine learning (ML) models. As part of a holistic testbed, we introduce a collection of reproducible tasks, as well as a toolchain that controls the dynamic migration of tasks depending on ECU status and load. The migration is aided by the machine learning predictions on the schedulability analysis of possible future task distributions. To demonstrate the capabilities of the setup, we show its integration with FreeRTOS-based ECUs and two ML models—a long short-term memory (LSTM) network and a spiking neural network—along with a collection of tasks to distribute among the ECUs. Our approach shows a promising potential for machine-learning-based schedulability analysis and enables a comparison between different ML models.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Ecu System Implementationmentioning

confidence: 99%

Section: Lstm Networkmentioning

confidence: 99%

Section: Machine Learningmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations