Detail View

Mitigating DDS Discovery Bursts in ROS 2 Multi-Robot System

Citations

WEB OF SCIENCE

Citations

SCOPUS

Metadata Downloads

DC Field Value Language
dc.contributor.advisor 박경준 -
dc.contributor.author Taehyun Kim -
dc.date.accessioned 2026-01-23T10:57:13Z -
dc.date.available 2026-01-24T06:00:40Z -
dc.date.issued 2026 -
dc.identifier.uri https://scholar.dgist.ac.kr/handle/20.500.11750/59728 -
dc.identifier.uri http://dgist.dcollection.net/common/orgView/200000942536 -
dc.description Robot Operating System 2, Data Distribution Service, Discovery Protocol, Multi node Multi Robot System, Wireless Network -
dc.description.abstract This paper analyzes the ROS 2 Data Distribution Service (DDS) discovery stack in multi-node wireless multi-robot systems (MRS), down to the link-layer software transmit queue (TxQ), and identifies three root causes of in-system packet drops: (i) the initial burst traffic of the Simple Participant Discovery Protocol (SPDP) and the Simple Endpoint Discovery Protocol (SEDP); (ii) the probabilistic dynamics of the Heartbeat–AckNack-based retransmission mechanism; and (iii) the interaction with the software TxQ waiting time threshold. To this end, we establish a software TxQ drop model that incorporates the initial payload and service slots, together with a retransmission-probability model based on the per-attempt success probability, p. Building on these models, we propose Staggered Pacing of ROS 2 Enablement and DDS (SPRED), an algorithm that computes node start times from the sizes of the initial SPDP/SEDP bursts and the software TxQ throughput to temporally disperse traffic. The approach is directly applicable to standard ROS 2 configurations without modifying lower-layer protocols.
Experimentally, using an IEEE 802.11n 2.4 GHz wireless network and a ROS 2 Humble system with Fast DDS 2.6.9, we estimated the model parameters and independently evaluated the SEDP retransmission probability in a wired setting. As a result, SPRED shortened discovery runtime by up to 35.5%, reduced the average number of transmitted packets by up to 22.6%, and lowered the average drop rate by 36.7%, 31.0%, and 33.3% for the 10, 12, and 14 node configurations, respectively. These gains result from temporally spreading the initial burst, which mitigates the concentration of control and data frames within the same 0.1-s window, reduces overlap among retransmissions during early convergence, and keeps the fraction of packets whose waiting time in the software TxQ exceeds 0.2 s low. We confirm that SPRED achieves robust discovery in wireless ROS 2 systems.|본 논문은 다중 노드 무선 멀티 로봇 시스템에서 로봇운영체제2의 발견 프로토콜 트래픽이 유발하는 병목과 그 근본 원인을 규명한다. 로봇운영체제2가 널리 쓰이지만 버스트형 제어 트래픽이 무선 링크와 하위 계층에 미치는 영향은 충분히 해명되지 않았다. 우리는 무선 환경에서 로봇운영체제2, 데이터 분산 서비스, 발견 프로토콜 그리고 링크 계층의 소프트웨어 전송 큐까지 분석해, (1) 발견 프로토콜에서의 과도한 초기 버스트 패킷 전송, (2) 성공확률 기반 재전송의 타이밍 중첩, (3) IEEE 802.11 소프트웨어 전송 큐의 대기시간 임계와의 상호작용으로 인한 패킷 드롭이라는 세 가지 핵심 문제를 규명한다. 이에 따라 초기 버스트 크기와 링크 처리량을 바탕으로 노드 시작 시점을 조정해 트래픽을 시간축으로 분산하는SPREAD 알고리즘을 제안한다. 제안 방식은 하위 프로토콜을 수정하지 않고 표준 로봇운영체제2 구성으로 적용 가능하며 분산성을 유지한다. 로봇운영체제2 험블–빠른 데이터 분산 서비스 2.6.9, IEEE 802.11n의 2.4 GHz 환경에서의 실험 결과, SPREAD는 기존 대비 발견 프로토콜 진행 시간을 최대 35.5% 단축하고 전송 패킷 수를 최대 22.6% 줄였으며, 평균 드롭률도 10, 12, 14노드에서 각각 36.7%, 31.0%, 33.3% 낮춰 무선 환경에서도 견고한 발견 프로토콜 성능을 달성함을 확인했다.
-
dc.description.tableofcontents Abstract
List of contents
List of algorithms
List of figures
INTRODUCION
RELATED WORK
PRELIMINARIES
ROS2 Transmission Stack
Data Distribution Service
Discovery Protocol
Simple Participant Discovery Protocol
Simple Endpoint Discovery Protocol
Quality of Service
RELIABLE
TRANSIENT LOCAL
SYSTEM MODEL AND SPREAD ALGORITHM
Simple Participant Discovery Protocol Model
Simple Endpoint Discovery Protocol Model
Software TxQ Drop Model
SPREAD Algorithm
PERFORMANCE EVALUATION
Experimental Configuration and Performance Indicators
Experimental Setup
Experimental Scenarios
Experimental Results and Observations
Software TxQ Throughput Results
Empirical Pr(n) probability Results
SPREAD algorithm Results
CONCLUSION
REFERENCES
SUMMARY(Korean)
-
dc.format.extent 25 -
dc.language eng -
dc.publisher DGIST -
dc.title Mitigating DDS Discovery Bursts in ROS 2 Multi-Robot System -
dc.title.alternative 로봇운영체제 2인 멀티 로봇 시스템에서 데이터 분산 서비스 발견 과정의 버스트 완화 -
dc.type Thesis -
dc.identifier.doi 10.22677/THESIS.200000942536 -
dc.description.degree Master -
dc.contributor.department Artificial Intelligence Major -
dc.date.awarded 2026-02-01 -
dc.publisher.location Daegu -
dc.description.database dCollection -
dc.citation XT.AM 김88 202602 -
dc.date.accepted 2026-01-19 -
dc.contributor.alternativeDepartment 학제학과인공지능전공 -
dc.subject.keyword Robot Operating System 2, Data Distribution Service, Discovery Protocol, Multi node Multi Robot System, Wireless Network -
dc.contributor.affiliatedAuthor Taehyun Kim -
dc.contributor.affiliatedAuthor Kyung-Joon Park -
dc.contributor.alternativeName 김태현 -
dc.contributor.alternativeName Kyung-Joon Park -
Show Simple Item Record

File Downloads

  • There are no files associated with this item.

공유

qrcode
공유하기

Total Views & Downloads

???jsp.display-item.statistics.view???: , ???jsp.display-item.statistics.download???: