On the suboptimality of a parallel Kalman filter

A filtering scheme has been proposed in the literature that attempts to achieve a certain amount of parallelism in a discrete-time Kalman filter by forcing the measurement- and time-update levels of the filter to decouple whereby the two levels may be run in parallel. The authors use an error covariance analysis to prove that the proposed filter is suboptimal and is not, in fact, a Kalman filter.<
>