Gokhan Soysal

An Adaptive Kalman Filter For Tracking Maneuvering Targets

In this paper, an adaptive Kalman filter is presented. The proposed filter calculates the process noise covariance which determines the tracking ability of the Kalman filter at every update time. Thus, the filter becomes sensitive to variations in the the target motion. In the filter, process noise covariance is updated at every sampling interval according to a predetermined relationship between the innovation covariance of the Kalman filter and available data form the measurements. Then state estimation and state estimation covariance are updated using the new process noise covariance. Tracking performance of the proposed algorithm has been compared to the Interactive multiple model filter through simulations

Feasibility of UMTS signal for passive radar

Passive radar systems are non-cooperative systems that can detect targets using illuminators already in place. Detection and tracking are carried out with lower costs, difficulty of jamming and including no need for frequency allocation by utilizing different types of waveforms such as television, radio, mobile communication signals. Along with the increase in UMTS network coverage, the suitability for passive systems has become significant. In this study, the ambiguity function was analyzed by collecting real data with the measurement platform to access the feasibility of these types of signals in passive radar systems and performance parameters showing that UMTS signal is suitable for passive radar system were presented.

Performance of the interacting multiple model filter under high measurement rates

Interacting Multiple Model (IMM) is one of the most successful techniques proposed for maneuvering target tracking. The method works as a filter bank where each model represents a possible target motion and knows how the other models perform. One of the performance parameters, perhaps the least attended, is the predetermined Markov chain transition matrix. In this paper we present a study on how to determine the elements of the matrix and support it with simulation results.

Analysis of observed information in multistatic radar network

In this study, observed information matrix has been calculated in a multistatic radar network where an Unscented Kalman filter, that utilized bistatic range and range rate measurements, was employed as a 3D tracker. Variation of the observed information with respect to targets location in the network has been analyzed through simulations and the results are presented in contrast to tracking error.