Fault Tolerance by Using Clustering Detection in Wireless Sensor and Actuator Networks

• Main Authors: Chien-Hsing Chung, 鍾見興
• Other Authors: Chiu-ching Tuan
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/4559zq

碩士 === 國立臺北科技大學 === 電腦與通訊研究所 === 98 === In Wireless Sensor and Actuator Networks (WSAN), when the power of sensor nodes is insufficient or wireless communications and deployment environment become unstable, the sensed data may lose in the way to the destination. Therefore, how to design an efficient fault tolerance mechanism becomes an important issue in WSAN. However, the existing fault-tolerant mechanisms are mostly concentrated on the routing link error handling. That is, research for error sensing data is little. When sensor nodes may sense inaccurately or invalidly, the error sensing data will cause the actuator to make error response and handling. Therefore, this paper proposed a Clustering Detection Method (CDM) based on WSAN. This method use the discrete degree of sensing data to judge the correctness by ruling out the error sensor nodes. CDM also can combine with routing link fault tolerant mechanism to construct a complete fault tolerance mechanism. In performance analysis, we compare CDM with Distributed Fault Detection Method (DFDM), and define three metrics: Faulty Sensor Detection Accuracy (FSDA), False Alarm Rate (FAR), Monitoring Scope and Sensor Nodes Relationship (MSSNR) for measuring the performance of CDM and DFDM. From the simulation results, when the average neighbor of nodes 7, CDM was better than DFDM by 28% on the FAR, by 37% on the FSDA, respectively. When the average neighbor of nodes is 20, CDM was better than DFDM by 10% on the FAR, by 19% on the FSDA, respectively. In scope of monitoring in 600×600 m2 and the fault rate of 25%, the hardware cost of CDM is only 20% for DFDM. In Scope of monitoring in 200×200 m2 and the fault rate of 25%, the hardware cost of CDM is only 17% of cost for DFDM.