Properties of a Robust Kalman Filter

AbstractAlmost all the communication and control systems may suffer from insufficient measurement data either due to sensor faults or communication errors. Standard Kalman filter predicts state of the system and then tunes that with the help of newly arrived observations. But in the case of insufficient data the question arises as how to compensate the loss of observation in the state estimation. In this paper a robust estimation design is presented for a sampled linear system where the sensor readings are subjected to random loss. Several easy-to-implement approaches are discussed in this paper. A brief description is stated on the design structure, affected state and minimum error covariance matrix. A comprehensive comparison survey for these approaches is presented which shows various features like computational time, innovation, convergence of the affected riccati equation, etc.