Multiagent systems consist of groups of agents that exchange local information to achieve given system-level goals. Traditionally, these systems assumed that all agents have some knowledge (e.g., observations of a process of interest) to contribute in order to achieve a common goal of interest, which may not be the case in many scenarios. Motivated from this standpoint, a new class of multiagent systems, active-passive multiagent systems, were recently proposed, which consists of agents subject to observations of a process of interest (i.e., active agents) and agents without any observations (i.e., passive agents). However, in this class of multiagent systems, agents need to continuously exchange information among themselves, incurring a high cost of interagent information exchange, and requiring that agents synchronize their update times. To address these shortcomings, this paper presents a new class of event-triggered active-passive dynamic consensus filters, in which agents schedule information exchange dependent on errors exceeding user-defined thresholds, significantly reducing the overall cost of interagent information exchange, and allowing agents to determine when to broadcast their information to their neighbors thus eliminating the need to synchronize their states. An illustrative numerical example is presented to demonstrate the efficacy of our approach.
- Bandpass filters,
- Information dissemination,
- Multi agent systems,
- Active agents,
- Consensus filter,
- Event-triggered,
- Information exchanges,
- Local information,
- Overall costs,
- Passive dynamics,
- System levels,
- Passive filters
Available at: http://works.bepress.com/jagannathan-sarangapani/193/