Multi-Model based Object Tracking Architecture with Model Selection Strategies for Wireless Sensor Network

• Main Authors: Chien-Han Liao, 廖千涵
• Other Authors: Chiang Lee
• Format: Others
• Language: en_US
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/19723272235918984281

碩士 === 國立成功大學 === 資訊工程學系碩博士班 === 96 === The tracking techniques play a significant role in wireless sensor network applications, such as troops monitoring and wildlife habitat monitoring. The methodology of recently proposed tracking schemes predicts the object location in the future based on a moving model, and then activate nearby sensors to monitor the target periodically. However, existing tracking systems cannot effectively track targets for a long time. This is because in most real-world applications, a target frequently or occasionally moves with different patterns, and accurately predicting the movement of a target needs multiple moving models, instead of a single model. Thus, the existing schemes frequently incur target loss, and a great amount of energy is consumed to find back the target. In this thesis, we propose a tracking framework, called Multi-Model based Objet Tracking Architecture (MMOTA), to energy-efficiently track a moving target. MMOTA can dynamically select the best tracking modules to monitor the target among the multiple predesigned tracking modules in different situations. Next, we derive a Monitoring-Cost Evaluator to evaluate the monitoring cost for the inactive tracking modules, and then design three tracking module selection strategies, including Greedy Strategy, Min-Max Strategy, and Weighted Moving Average Strategy, to select the most effectively tracking module for monitoring the target. Finally, we conduct a set of comprehensive experiments to compare MMOTA against the existing tracking systems and evaluate the three proposed tracking module selection strategies. The result shows that MMOTA consumes the much less energy for monitoring the target than the existing tracking systems, and saves more than 50.7% amount of energy for monitoring the target.