In Press                   Back to the articles list | Back to browse issues page

XML Print

An Experimental Analysis of the Latency of Linux Kernels Applicable for Real-Time Control Strategies
Ayoub Khodaparast , Hassan Ghiti Sarand
Abstract:   (61 Views)
Real-time control applications, crucial in robotics, industrial automation, and medical devices, demand precise and predictable timing for reliable operation. This paper presents an experimental investigation into the latency performance of various Linux kernels, including standard Linux, a low-latency kernel, Xenomai, and a real-time kernel patched with PREEMPT_RT. Our test setup utilizes a data acquisition card to measure the latency between sending and receiving a pulse signal through analog input-output channels, generated by a C++ code. This latency metric serves as an indicator of the responsiveness of the kernel and other control objects on a specific computer system. Our experiments were conducted under a wide range of conditions to comprehensively assess latency performance. This includes different versions of standard and real-time Linux kernels, varying numbers of CPU cores, program priority levels, data saving rates, a range of data acquisition cards, communication protocols, thread assignments to processor cores, and test durations. The results highlight the importance of long-term testing to accurately determine the maximum latency. Furthermore, the findings demonstrate significantly lower latency for the PREEMPT_RT patched kernel across various tests, indicating its suitability for demanding real-time control applications that require tight timing constraints.
Keywords: Latency measurement, Real-time control, Real-time Linux, Linux kernel
Full-Text [PDF 1497 kb]   (36 Downloads)    
Type of Study: Research Paper | Subject: Deep Learning
Received: 2025/07/08 | Revised: 2025/12/11 | Accepted: 2025/09/06
Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Creative Commons License
© 2022 by the authors. Licensee IUST, Tehran, Iran. This is an open access journal distributed under the terms and conditions of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) license.