Estimation of Parameters in Avian Movement Models

TL;DR

This paper introduces a novel parameter estimation method for avian movement models using state space modeling, extended Kalman filter, and maximum likelihood estimation, validated through a kite simulation.

Contribution

It proposes a new approach to estimate unknown parameters in bird movement models with improved accuracy using unscented Kalman filtering.

Findings

Effective parameter estimation demonstrated with a kite simulation.

Enhanced accuracy in state estimation via unscented transformation.

Method provides reliable movement tracking for ecological studies.

Abstract

The knowledge of the movement of animals is important and necessary for ecologists to do further analysis such as exploring the animal migration route. A novel method which is based on the state space modeling has been proposed to track the bird, where the VHF transmitter is attached to the bird to emit the signal and several towers with antenna arrays installed on its top are built to receive the signal. The method consists of two parts, the first one is called movement model which accounts for prediction of the dynamic movement of the target, and the second part is the measurement model which links the target's state variables to the available measurements data, the measurement includes the time when the signal was detected, the ID of the antenna array which detected the signal and integers between 0 and 255, the integers are proportional to the strength of received signal. The extended Kalman filter is then applied to estimate the location of the target with combing the movement model and measurement model. In the movement model, several parameters with positive values are deployed to define the change of the state variables with time, these parameters reflect the relationship of the state variables at current time and next time. In this paper, a method based on the maximum likelihood estimation is proposed to estimate the appropriate values for these unknown constant variables with given measurement data, and a kite is applied to demonstrate the validity of the proposed method. Furthermore, the unscented transformation is applied in Kalman filter to achieve more accurate estimation of the target's states.