TY - JOUR
T1 - Determining Redundant Links of Multiagent Systems in Keeping or Improving Consensus Convergence Rates
AU - Chen, Xinzhuang
AU - Zhang, Shenggui
AU - Zhang, Lijun
AU - Yu, Guihai
AU - Liu, Jiuqiang
N1 - Publisher Copyright:
© 2007-2012 IEEE.
PY - 2022/12/1
Y1 - 2022/12/1
N2 - For a multiagent system (MAS), some redundant links can be removed without reducing its convergence rate. The deletion of redundant links is of benefit to an MAS in improving its lifetime, resources usage efficiency, and so on. In this article, the problem of determining redundant links of MASs is addressed. Based on the theory of graph spectra, necessary and sufficient conditions for determining a redundant link are given for an MAS under the first-order and the second-order consensus protocols. It is surprising to find that the convergence rate of the second-order protocol can even be improved by removing some redundant links, which can be determined by the proposed condition. Moreover, algorithms with running time $O(N^5)$ are designed to determine a maximal set of redundant links for MASs with $N$ agents. Numerical simulations are given to illustrate the effectiveness of the algorithms.
AB - For a multiagent system (MAS), some redundant links can be removed without reducing its convergence rate. The deletion of redundant links is of benefit to an MAS in improving its lifetime, resources usage efficiency, and so on. In this article, the problem of determining redundant links of MASs is addressed. Based on the theory of graph spectra, necessary and sufficient conditions for determining a redundant link are given for an MAS under the first-order and the second-order consensus protocols. It is surprising to find that the convergence rate of the second-order protocol can even be improved by removing some redundant links, which can be determined by the proposed condition. Moreover, algorithms with running time $O(N^5)$ are designed to determine a maximal set of redundant links for MASs with $N$ agents. Numerical simulations are given to illustrate the effectiveness of the algorithms.
KW - Convergence rate
KW - Laplacian eigenvalues
KW - multiagent systems (MASs)
KW - redundant links
UR - http://www.scopus.com/inward/record.url?scp=85121834946&partnerID=8YFLogxK
U2 - 10.1109/JSYST.2021.3130591
DO - 10.1109/JSYST.2021.3130591
M3 - 文章
AN - SCOPUS:85121834946
SN - 1932-8184
VL - 16
SP - 6153
EP - 6163
JO - IEEE Systems Journal
JF - IEEE Systems Journal
IS - 4
ER -