Parallel Implementation of M-Step Kalman FIR Filter for Linear Discrete Time-invariant Systems

This paper proposes a new version of the receding horizon Kalman FIR (RHKF) filter called M-step Kalman FIR filter. The proposed filter is with an FIR filter form and can be represented in an iterative form as well as in a standard FIR form. The proposed filter can be implemented in parallel which is call parallel M-step Kalman FIR filter and thus the computation time can be reduced compared with that of the existing RHKF filter while it still preserves the good properties of the RHKF filter such as deadbeat and unbiasedness. The validity of the proposed filter is illustrated by simulation studies