MILLÁN GATA, PABLO, ORIHUELA ESPINA, DIEGO LUIS, JURADO FLORES, ISABEL, Vivas, Carlos , Rubio, Francisco R.
No
Int. J. Syst. Sci.
Article
Científica
1.579
0.758
07/03/2013
000344245000010
2-s2.0-84908055693
This paper's aim is to present a novel design technique for distributed estimation in networked systems. The problem assumes a network of interconnected agents each one having partial access to measurements from a linear plant and broadcasting their estimations to their neighbours. The objective is to reach a reliable estimation of the plant state from every agent location. The observer's structure implemented in each agent is based on local Luenberger-like observers in combination with consensus strategies. The paper focuses on the following network related issues: delays, packet dropouts and communication policy (time and event-driven). The design problem is solved via linear matrix inequalities and stability proofs are provided. The technique is of application for sensor networks and large scale systems where centralized estimation schemes are not advisable and energy-aware implementations are of interest. Simulation examples are provided to show the performance of the proposed methodologies.
event-based sampling; packet dropouts; network-induced delays; sensor networks; distributed estimation