Coordinated One-Step Optimal Distributed State Prediction for a Networked Dynamical System

Distributed observer design for networked dynamical systems

“…Consider the networked system Σ like the one investigated in [11] and modified from [15], which consists of LTI dynamic subsystems and the dynamics of its -th subsystem Σ is described by the following state space model like representation,…”

“…It is assumed throughout this paper that the dimensions of the vectors ( , ), ( , ), ( , ) and ( , ), are respectively x , v , z and y . Without loss of generality, it is assumed that every row of the matrix Φ has only one nonzero element which is equal to one [11]. From [16] …”

“…However, as argued in [11] and [19], although the matrices * # , * , # = T or S are block diagonal and subsystems connection matrix Φ is sparse, the state transition matrix is usually dense. When the networked dynamical systems have a large number of subsystems, the computation of stability analysis may meet prohibitive implementation difficulties.…”

“…Under these situations, direct application of observer design methods based on linear systems may usually encounter prohibitive implementation difficulties. Therefore, distributed observer designs using only local system output measurements are generally much more appreciated [7][8][9][10][11][12]. In [9], a distributed Kriged Kalman filter is designed for predictive inference of a robotic sensor network and of its gradient.…”

“…In [10], the well known Kalman filter is extended to a multiple dimensional Roesser model utilizing the concept of wave advance process. In [11], a recursive one-step state prediction procedure based only on local system output measurements is derived for a networked dynamic system. The estimator can be easily realized in a distributed way, and can be scaled to systems with a large quantity of subsystems.…”