Sensor networks comprise a group of agents equipped with sensing devices and communicating capabilities in order to solve the common task of cooperative estimation of a detectable physical process. In this framework, each agent in the system activates, in a distributed fashion, an estimator, which relies on local measurements fused with the estimates from other agents in the network.
A recently developed tool to solve this problem is the introduction of a consensus-based term fused with a classical state estimator structure, knows as the consensus Kalman filter. In this architecture, each estimator incorporates a classic Kalman term, along with a consensus term used to fuse the estimates from neighboring agents. Our contribution begins with proposing a method based on semi-definite programming to compute a centralized consensus gain term leading to improved performance of the estimator over existing solutions.
We also propose a decentralized consensus gain that can be computed by each agent and relies only on local network properties.
We further extend our research to tackle the important aspects of reducing energy (communication) consumption in network applications. To do so, we utilize an event triggering mechanism in which communication is permitted only if certain conditions are met. The main analytical challenge in these estimator structures is the design of the consensus gain term and an event trigger condition that ensures stability while maintaining a prescribed degree of performance. In this direction, our contribution continues with proposing both a centralized and a decentralized consensus gain along with a tailored event triggered condition. We show that these event trigger estimators out-performs the standard non-cooperative local Kalman filter. We provide numerical simulations to demonstrate the effectiveness of our results compared to existing solutions in the literature.
