FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2918946021> ?p ?o ?g. }

• W2918946021 abstract "Nowadays, optimization is a pervasive tool, employed in a lot different fields. Due to its flexibility, it can be used to solve many diverse problems, some of which do not seem to require an optimization framework. As so, the research on this topic is always active and copious. Another very interesting and current investigation field involves multi-agentsystems, that is, systems composed by a lot of (possibly different) agents. The research on cyber-physical systems, believed to be one of the challenges of the 21st century, is very extensive, and comprises very complex systems like smart cities and smart power-grids, but also much more simple ones, like wireless sensor networks or camera networks. In amulti-agent context, the optimization framework is extensively used. As a consequence, optimization in multi-agent systems is an attractive topic to investigate. The contents of this thesis focus on distributed optimization within a multi-agent scenario, i.e., optimization performed by a set of peers, among which there is no leader. Accordingly, when these agents have to perform a task, formulated as an optimization problem, they have to collaborate to solve it, all using the same kind of update rule.Collaboration clearly implies the need of messages exchange among the agents, and the focus of the thesis is on the criticalities related to the communication step. In particular, no reliability of this step is assumed, meaning that the packets exchanged between two agents can sometime be lost. Also, the sought-for solution does not have to employ an acknowledge protocol, that is, when an agent has to send a packet, it just sends it and goes on with its computation, without waiting for a confirmation that the receiver has actually received the packet. Almost all works in the existing literature deal with packet losses employing an acknowledge (ACK) system; the effort in this thesis is to avoid the use of an ACK system, since it can slow down the communication step. However, this choice of averting the use of ACKs makes the development of optimization algorithms, and especially their convergence proof, more involved. Apart from robustness to packet losses, the algorithms developed in this dissertation are also asynchronous, that is, the agents do not need to be synchronized to perform the update and communication steps.Three types of optimization problems are analyzed in the thesis. The first one is the patrolling problem for camera networks. The algorithm developed to solve this problem has a restricted applicability, since it is very task-dependent. The other two problems are more general, because both concern the minimization of the sum of cost functions, one for each agent in the system. In the first case, the form of the local cost functions is particular: these, in fact, are locally coupled, in the sense that the cost function of an agent depends on the variables of the agent itself and on those of its direct neighbors. The sought-for algorithm has to satisfy two properties (apart from asynchronicity and robustness to packet losses): the requirement of asking a single communication exchange per iteration (which also reduces the need of synchronicity) and the requirement that the communication among agents is only between direct neighbors. In the second case, the local functions depend all on the same variables. The analysis first focuses on the special case of local quadratic cost functions and their strong relationship with the consensus problem. Besides the development of a robust and asynchronous algorithm for the average consensus problem, a comparison among algorithms to solve the minimization of the sum of quadratic cost functions is carried out. Finally, the distributed minimization of the sum of more general local cost functions is tackled, leading to the development of a robust version of the Newton-Raphson consensus.The theoretical tools employed in the thesis to prove convergence of the algorithms mainly rely on Lyapunov theory and the separation of scales theory." @default.
• W2918946021 created "2019-03-11" @default.
• W2918946021 creator A5047980919 @default.
• W2918946021 date "2018-01-14" @default.
• W2918946021 modified "2023-09-24" @default.
• W2918946021 title "Multi-Agent Distributed Optimization and Estimation over Lossy Networks" @default.
• W2918946021 hasPublicationYear "2018" @default.
• W2918946021 type Work @default.
• W2918946021 sameAs 2918946021 @default.
• W2918946021 citedByCount "0" @default.
• W2918946021 crossrefType "journal-article" @default.
• W2918946021 hasAuthorship W2918946021A5047980919 @default.
• W2918946021 hasConcept C11413529 @default.
• W2918946021 hasConcept C120314980 @default.
• W2918946021 hasConcept C120665830 @default.
• W2918946021 hasConcept C121332964 @default.
• W2918946021 hasConcept C137836250 @default.
• W2918946021 hasConcept C151730666 @default.
• W2918946021 hasConcept C154945302 @default.
• W2918946021 hasConcept C177264268 @default.
• W2918946021 hasConcept C192209626 @default.
• W2918946021 hasConcept C199360897 @default.
• W2918946021 hasConcept C2779343474 @default.
• W2918946021 hasConcept C41008148 @default.
• W2918946021 hasConcept C41550386 @default.
• W2918946021 hasConcept C86803240 @default.
• W2918946021 hasConceptScore W2918946021C11413529 @default.
• W2918946021 hasConceptScore W2918946021C120314980 @default.
• W2918946021 hasConceptScore W2918946021C120665830 @default.
• W2918946021 hasConceptScore W2918946021C121332964 @default.
• W2918946021 hasConceptScore W2918946021C137836250 @default.
• W2918946021 hasConceptScore W2918946021C151730666 @default.
• W2918946021 hasConceptScore W2918946021C154945302 @default.
• W2918946021 hasConceptScore W2918946021C177264268 @default.
• W2918946021 hasConceptScore W2918946021C192209626 @default.
• W2918946021 hasConceptScore W2918946021C199360897 @default.
• W2918946021 hasConceptScore W2918946021C2779343474 @default.
• W2918946021 hasConceptScore W2918946021C41008148 @default.
• W2918946021 hasConceptScore W2918946021C41550386 @default.
• W2918946021 hasConceptScore W2918946021C86803240 @default.
• W2918946021 hasLocation W29189460211 @default.
• W2918946021 hasOpenAccess W2918946021 @default.
• W2918946021 hasPrimaryLocation W29189460211 @default.
• W2918946021 hasRelatedWork W1563575792 @default.
• W2918946021 hasRelatedWork W1861342261 @default.
• W2918946021 hasRelatedWork W2167462080 @default.
• W2918946021 hasRelatedWork W2185110050 @default.
• W2918946021 hasRelatedWork W23560277 @default.
• W2918946021 hasRelatedWork W2737312012 @default.
• W2918946021 hasRelatedWork W2757445497 @default.
• W2918946021 hasRelatedWork W2892781498 @default.
• W2918946021 hasRelatedWork W2898963448 @default.
• W2918946021 hasRelatedWork W2945819093 @default.
• W2918946021 hasRelatedWork W2947662456 @default.
• W2918946021 hasRelatedWork W2993044575 @default.
• W2918946021 hasRelatedWork W2994937904 @default.
• W2918946021 hasRelatedWork W3007472775 @default.
• W2918946021 hasRelatedWork W3007832921 @default.
• W2918946021 hasRelatedWork W3047059968 @default.
• W2918946021 hasRelatedWork W3103567921 @default.
• W2918946021 hasRelatedWork W3116610253 @default.
• W2918946021 hasRelatedWork W3151046600 @default.
• W2918946021 hasRelatedWork W66329308 @default.
• W2918946021 isParatext "false" @default.
• W2918946021 isRetracted "false" @default.
• W2918946021 magId "2918946021" @default.
• W2918946021 workType "article" @default.