□ A team led by Prof. Park Kyung-joon from the Department of Electrical Engineering and Computer Science at DGIST (President Kunwoo Lee) has developed the world's first mathematical model to quantitatively analyze the data delivery latency in the Data Distribution Service (DDS)[1] within the Robot Operating System (ROS) 2[2]. The research team expressed the time delay occurring in the communication process in a closed form, accounting for changes in key communication parameters. This lays the foundation for the systematic analysis of the communication delay problem, which has previously been difficult to address.
□ Currently, various robot software frameworks are being developed, but the ROS is attracting attention due to its steady progress and active community support. While the original ROS 1 used Transmission Control Protocol (TCP)[3]-based communication, the ROS 2, released in 2017, adopted the DDS as its middleware and changed to User Datagram Protocol (UDP)[4]-based communication. This change allows for high scalability and flexibility. In particular, data are delivered reliably in the DDS as they are retransmitted at the application layer, unlike in TCP, resulting in different delay characteristics.
□ To quantitatively evaluate the impact of the DDS structure on communication delay, Prof. Park's team proposed a time delay model that considers key communication parameters, such as packet delivery ratio, data period, and heartbeat period. The team experimentally validated the proposed model using 35 different scenarios and found that its results matched the actual measurements, with an average error range of 6.88%. These findings are expected to provide design guidelines for minimizing communication delays in ROS 2 applications.
□ "This is the first study to specifically investigate the delay in the reliability delivery structure for the ROS 2 DDS from a network perspective," said Prof. Park from the Department of Electrical Engineering and Computer Science at DGIST. "The performance of a robot can vary significantly depending on how key communication parameters, such as data period and heartbeat period, are set. We expect this research to serve as an important guide for efficient communication design in the fields of robotics and industrial automation in the future."
□ The research was supported by the Unmanned Swarm CPS Research Lab Program of the Defense Acquisition Program Administration and the Agency for Defense Development, as well as the Mid-Career Researcher Support Program of the National Research Foundation of Korea. The research findings will be presented at IEEE INFOCOM 2025, a leading international conference in the field, to be held in London in May. The first author, Park Hyung-seok, a Ph.D. student, has founded S-Innovations, a robotics software startup, as a student-led initiative to drive the commercialization of the technology.
- Corresponding Author E-mail Address : [email protected]
[1] Data Distribution Service (DDS): A standardized middleware protocol designed to enable reliable, real-time data exchange between multiple devices or systems in a networked environment.
[2] ROS 2: An open-source operating system for robotics software development that enables real-time data processing and reliable communications in a variety of environments.
[3] TCP: Internet protocol that operates in a connection-oriented manner to ensure reliable and ordered data transfers.
[4] UDP: A connectionless protocol that transfers data quickly without establishing a connection, prioritizing speed and efficiency over reliability.
