| Summary: | 碩士 === 國立臺灣科技大學 === 資訊工程系 === 99 === Wireless sensor networks have been used for a variety of purposes such as wildlife habitats monitoring, malicious enemy detection, and climate observation. Conventionally, stationary sensors are used to carry out sensing tasks. However, stationary sensors could lead to many problems such as communication overhead and coverage holes.
This thesis investigates the problem of detecting a target or event using mobile sensors in a region with or without obstacles. The goal is to maximize the worst-case detection probability over the monitored region and reduce patrolling length of mobile sensors. An approach, namely critical sensing location path planning (CSLP), is developed to guide mobile sensors in order to collect data in an efficient way. CSLP directs mobile sensors to take measurements at carefully selected locations and reduces the patrolling path by solving a Travelling Salesman Problem. The proposed method can be used for cases either in the presence or absence of obstacles. Moreover, the thesis considers load balance problem if multiple mobile sensors are used. A method based on K-means clustering algorithm is developed to balance patrolling load for each sensor.
The performance of CSLP is shown by simulations for regions with and without obstacles. The results show that CSLP is an effective solution for the proposed mobile sensing architecture in terms of detection probability and patrolling path length of mobile sensors. In addition, the performance of the K-means clustering based balance algorithm is compared to one based on a minMax clustering algorithm. Simulation results show that the former outperforms the later in balancing patrolling length for each sensor.
|
|---|
