Nonlinear ltering of dynamical systems is an extensively studied problem, usually tackled extending the well-established optimality results existing for linear systems, i.e. the Kalman lter, in the realm of nonlinear functions. The approaches are numerous [1], starting from the early developed linearized Kalman lter and extended Kalman lter (EKF), soon rened to attain better performance in case of severe nonlinearities. Some methods retain higher order derivatives as with the second order EKF, others rely on iterative linearization. This last concept was most often applied to the measurement equation, leading to the iterated extended Kalman lter (IEKF), which improves upon the EKF by iterating the nonlinear measurement update equation through a re-linearization about the updated state estimate, while retaining the time update step equal to the EKF one. The application of the re-linearization technique to handle also nonlinear dynamics received apparently less attention. An early example is represented by a single stage iterated lter/smoother (SSIFS), proposed by Wishner et al. in [2], which embeds a smoothing step within the lter to improve the past estimate when a new measurement is processed. The smoothed estimate is then used as a starting point to re-linearize the prediction step for the ltering stage. This concept was more recently generalized and merged with a moving window batch lter by Psiaki with his backward smoothing extended Kalman lter (BSEKF) [3], allowing the smoothing process to run into the past for more than one stage. Such a combined lter-smoother was shown to outperform both the EKF and the unscented Kalman lter (UKF) in a spacecraft attitude estimation problem. This