Finite-horizon resilient state estimation for complex networks with integral measurements from partial nodes

This paper proposes a finite-horizon state estimation method for a kind of complex network that suffers from randomly occurring gain variations. The method involves utilizing integral measurements from a portion of nodes in such complex networks. Integral measurements are carried out to characterize time delays that occur in signal acquisition together with real-time signal processing. Measurements from only partial nodes reflect the fact that signals of several sensors are unacquirable. A Gaussian random variable is utilized to depict the random appearance of gain variations during the practical implementation of estimators. The aim of this paper is to construct finite-horizon resilient estimators for complex networks in view of integral measurements from a portion of nodes that fulfill the specified H-infinity performance demand involving a specified disturbance attenuation level. Necessary and sufficient conditions are put forward to ensure that such ideal estimators exist by employing stochastic analysis as well as using the completing squares method. The gain parameters of the finite-horizon estimators are expressed by adopting the Moore-Penrose pseudoinverse and acquired through solving the solutions to a group of coupled backward recursive Riccati difference equations with constraint conditions. A confirmatory instance is carried out that demonstrates the feasibility of the newly developed estimation algorithm.