Parameter identification of discrete-time linear stochastic systems based on decentralized square-root information filtering

The paper proposes a new method for identifying parameters of discrete-time linear stochastic systems using decentralized square-root information filtering (DSRIF). The main contribution of the paper is the derivation of a new identification criterion formulated in terms of DSRIF outputs, such as square roots of information matrices and corresponding estimates of information vectors. An algorithm for its computation is also provided, which uses J-orthogonal transformations at the communication and assimilation stage for updating filter quantities. The method is validated through a numerical example of circular motion tracking with various configurations of measurement models. Simulations show accurate parameter identification, with improved precision as the number of sensors increases, especially when using sensors measuring the full state vector. This work establishes a unified framework for decentralized square-root information filtering and parameter identification, suitable for real-life applications in fault-tolerant control, environmental monitoring, and adaptive signal processing.