Multi-sensor estimation using CKF and DKF

This paper compares two methods of multi-sensor estimation using Kalman filter in a sensor network. We focused on group-sensor method and track-to-track fusion approach, as representatives of typical centralized and decentralized architectures, to obtain a final estimation for the states in an instant sensor network. Each strategy is clearly illustrated and simulated for a simple network containing two sensors. The results of estimations in each one presented in two distinct situations; the conditions in those the two sensors benefit from either same or different measurement noise variances. The MSE values as a criterion for assessing the error between the estimations´ and the states´ real values are obtained using a procedure containing the Monte-Carlo technique.