A decentralized square root information filter/smootherA number of developments has recently led to a considerable interest in the decentralization of linear least squares estimators. The developments are partly related to the impending emergence of VLSI technology, the realization of parallel processing, and the need for algorithmic ways to speed the solution of dynamically decoupled, high dimensional estimation problems. A new method is presented for combining Square Root Information Filters (SRIF) estimates obtained from independent data sets. The new method involves an orthogonal transformation, and an information matrix filter 'homework' problem discussed by Schweppe (1973) is generalized. The employed SRIF orthogonal transformation methodology has been described by Bierman (1977).
Bierman, G. J. (Factorized Estimation Applications, Inc. Sherman Oaks, CA, United States)
Belzer, M. R. (Business and Technological Systems, Inc. Seabrook, MD, United States)