Fuzzy inference system as a collaboration logic and a new M/N-based track initiation logic for wireless sensor networks

Abstract

According to the advances in digital electronics, micro-electro-mechanical systems and wireless communication technologies, it is viable to integrate a device with sensing, computing and wireless communication capabilities within tiny dimensions. In this thesis, we focus on the target tracking issues on wireless sensor networks. First, we try to generate a new collaboration logic via a fuzzy inference system. We utilize Euclidean distance and mutual information metrics as the fuzzy membership functions to achieve a better sensor collaboration measure in order to lead most informative sen- sor node to broadcast its information. Our second aim is to improve the performance of the track initiation process of a target tracking system in cluttered environments via a new M/N-based track initiation logic. In generic M/N logic, the consecutive detections inside the gate are considered with identical weights and the initiation counter is increased by one. However, if the newly detected observations in the gate are away from the center of the predicted target position, the observations may not be a real target, the origin of the measurement may be the clutter. Thus, we should adjust the initiation counter weight of the observation according to the distance measure while utilizing M/N logic for the track initiation decision. For this purpose, we envisage different weighting schemes to determine the initiation counter weighting value. Consequently, elliptical weighting scheme for the M/N track initiation logic shows promis- ing results for cluttered environments in terms of decreasing the false track initiations while sustaining an admissible level of true track initiations.