A sequential approach for distributed set-membership estimation to exploit redundant information.

This manuscript proposes a novel approach for the distributed set-membership estimation of linear time-invariant plants. The method is based on sequentially exploiting the information that the agents receive from their neighbourhood in an adequate manner, so that redundant information is not discarded, but used to further shrink the sets. To do so, they need to build the innovation-routing table, which integrates routing information and concepts of subspace observability. Then, sequential filtering steps are performed over the sets that contain the different modes in which the state vector is locally decomposed. The method is presented for sets described as zonotopes and takes advantage of the developments of well-known centralized zonotopic observers. Simulation results show that the use of redundant information reduces the uncertainty of the estimation, and the increment of computational cost is minor. Furthermore, the sensitivity of the method with respect to different parameters is analysed with numerical simulations and compared to similar estimators in the literature.

Distributed Estimation Techniques for Cyber-Physical Systems: A Systematic Review.

This paper undertakes a systematic review (SR) on distributed estimation techniques applied to cyber-physical systems (CPS). Even though SRs are not the common way to survey a theme in the control community, they provide a rigorous, robust and objective formula that should not be always ignored. The presented SR incorporates and adapts the guidelines recommended in other fields (mainly biosciences and computer sciences) to the field of automation and control and presents a brief description of the different phases that constitute an SR. As a result, this review compares the different techniques found in the literature in terms of: The proposed estimator (Kalman filter, Luenberger observer, Bayesian filter, etc.), the particular application within CPS, the design of the estimators (decentralized vs centralized), the amount of data required for implementation or the inclusion of experiments/simulations in the studies. Particular attention is paid to those papers that present some results in applications that include humans, animals or biological systems.

Data Fusion Based on Subspace Decomposition for Distributed State Estimation in Multi-Hop Networks.

This paper deals with the problem of estimating the distributed states of a plant using a set of interconnected agents. Each of these agents must perform a real-time monitoring of the plant state, counting on the measurements of local plant outputs and on the exchange of information with the rest of the network. These inter-agent communications take place within a multi-hop network. Therefore, the transmitted information suffers a delay that depends on the position of the sender and receiver in a communication graph. Without loss of generality, it is considered that the transmission rate and the plant sampling rate are both identical. The paper presents a novel data-fusion-based observer structure based on subspace decomposition, and addresses two main subproblems: the observer design to stabilize the estimation error, and an optimal observer design to minimize the estimation uncertainties when plant disturbances and measurements noises come into play. The performance of the proposed design is tested in simulation.