“…However, several limitations, including smoothness requirement of function, lack of convergence for highly non-linear systems are associated with the EKF. To circumvent the problems, several derivative-free methods such as the unscented Kalman filter (UKF) [3,4] and its variants [5,6], the Gauss-Hermite filter [7][8][9], the sparse grid Gauss-Hermite filter [10,11], the central difference filter [12], the cubature Kalman filter (CKF) [13] and its extensions [14,15], the high-degree cubature quadrature (CQ) Kalman filter [16,17] have been introduced. In all the above-mentioned filters, the prior and the posterior probability density functions (pdfs) are approximated as Gaussian and characterised by mean and covariance.…”