TY - GEN
T1 - Distributed topology control of dynamic networks
AU - Zavlanos, Michael M.
AU - Tahbaz-Salehi, Alireza
AU - Jadbabaie, Ali
AU - Pappas, George J.
PY - 2008
Y1 - 2008
N2 - In this paper, we present a distributed control framework for controlling the topology of dynamic multi-agent networks. Agents are equipped with local sensing and wireless communication capabilities, however, due to power constraints, they are required to switch between two modes of operation, namely active and sleep. The control objective investigated in this paper is to determine distributed coordination protocols that regulate switching between the operation modes of every agent such that the overall network guarantees multi-hop communication links among a subset of so called boundary agents. In the proposed framework, coordination is based on a virtual market where every request to switch off is associated with a bid. Combinations of requests are verified with respect to connectivity and the one corresponding to the highest aggregate bid is finally served. Other than nearest neighbor information, our approach assumes no knowledge of the network topology, while verification of connectivity relies on notions of algebraic graph theory as well as gossip algorithms run over the network. Integration of the individual controllers results in an asynchronous networked control system for which we show that it satisfies the connectivity specification almost surely. We finally illustrate efficiency of our scalable approach in nontrivial computer simulations.
AB - In this paper, we present a distributed control framework for controlling the topology of dynamic multi-agent networks. Agents are equipped with local sensing and wireless communication capabilities, however, due to power constraints, they are required to switch between two modes of operation, namely active and sleep. The control objective investigated in this paper is to determine distributed coordination protocols that regulate switching between the operation modes of every agent such that the overall network guarantees multi-hop communication links among a subset of so called boundary agents. In the proposed framework, coordination is based on a virtual market where every request to switch off is associated with a bid. Combinations of requests are verified with respect to connectivity and the one corresponding to the highest aggregate bid is finally served. Other than nearest neighbor information, our approach assumes no knowledge of the network topology, while verification of connectivity relies on notions of algebraic graph theory as well as gossip algorithms run over the network. Integration of the individual controllers results in an asynchronous networked control system for which we show that it satisfies the connectivity specification almost surely. We finally illustrate efficiency of our scalable approach in nontrivial computer simulations.
UR - http://www.scopus.com/inward/record.url?scp=52449108710&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=52449108710&partnerID=8YFLogxK
U2 - 10.1109/ACC.2008.4586894
DO - 10.1109/ACC.2008.4586894
M3 - Conference contribution
AN - SCOPUS:52449108710
SN - 9781424420797
T3 - Proceedings of the American Control Conference
SP - 2660
EP - 2665
BT - 2008 American Control Conference, ACC
T2 - 2008 American Control Conference, ACC
Y2 - 11 June 2008 through 13 June 2008
ER -