Innovation-based adaptive Kalman filter derivation

Abstract

In this paper we develop innovation-based adaptive Kalman filter (IAKF) based on linear time-varying (LTV) state-space model. Thorough argumentation of process and measurement noise covariance adaptation based on innovation sequence covariance is provided. As a result of implementing the proposed methods the process noise and measurement noise covariance matrices are adapted within a sliding window of a fixed depth.