Cognitive radar differs from traditional radar in that it develops behavioral rules in a self-organizing manner, benefiting from continuous interaction with the environment. Although the range and angle measurement errors vary depending on the return signal from the targets and environmental variables, typical tracking algorithms (such as conventional EKF) assume constant observation noise. One of the techniques described in the articles is adaptive updating of the EKF filter measurement noise covariance matrix, which uses endogenous data from the cognitive radar to investigate range and angle measurement errors and the effect of the received SNR. When the measurement noise covariance matrix is updated, the computational load increases. In this paper, we present a novel optimal technique utilizing mathematical relationship analysis to reduce unnecessary calculations. We simulated our method's performance in a tracking problem and compared its accuracy and computational load to traditional techniques. The simulation results show that in the new tracking algorithm based on cognitive radar, compared with the conventional EKF method while controlling the computational complexity, the tracking errors have significantly improved. When compared with the SNR-EKF algorithm, with an error increase of 4%, the volume of added calculations has been reduced by 22%.