Fuzzy clustering means data association algorithm using an adaptive neuro-fuzzy network

A significant problem in multi-sensor multi-target tracking system is measurement to track association. Based on fuzzy clustering means algorithm, an efficient algorithm has been proposed to solve this problem. The fuzzy clustering means data association (FCMDA) algorithm has better performance than the other already known fuzzy logic data association algorithms. However, it is still worthy to investigate the characteristics of the FCMDA algorithm, which has high accuracy in measurement to track association when targets are far from each other, while it has low accuracy when targets are close to each other. The FCMDA algorithm usually loses its performance in this situation, especially when the noise of measurement is high. In this paper, to overcome the disadvantage of the FCMDA algorithm, an adaptive neuro-fuzzy inference system (ANFIS) is used. The ANFIS adjusts the predicted state of targets which are used as cluster centers in the FCMDA algorithm. The ANFIS has the advantage of expert knowledge of fuzzy inference system and the learning capability of neural networks. This is so, since a trained ANFIS is able to compensate the effect of wrong data association in the FCMDA algorithm. Monte Carlo simulation results show considerable improvement in terms of accuracy and performance achieved by using the ANFIS in the FCMDA algorithm.