Technology Research Of Energy Efficient Target Detection Based On Wireless Sensor Networks

Target detection based on wireless sensor networks (WSNs) is one of the most extensive and important applications, and also one of the current research focuses. With adaptability, flexibility, low power consumption and other characteristics, target detection based on wireless sensor networks has replaced traditional monitoring methods such like radar monitoring, satellite monitoring in many practical applications. It is involved with the military, commercial, ecological, social life and many other fields, so target detection has very high significance of reality research.In the applications of target detection based on wireless sensor networks, energy efficiency and detection accuracy is the main two technical indicators. Since the energy is limited in wireless sensor networks, energy-efficient detection technology has always been the research focus. We could save energy by either adding sleep mechanism in the MAC protocol or designing certain network structure to reduce the unnecessary computation and data transfer. The detection accuracy includes alarm accuracy and false alarm rate, making alarm while detecting the exception and but avoiding the waste of resources of excessive false alarms. With different scenarios, target detection technology research focuses are different.According to the node configuration in the scenarios, we can classify the scenarios into two categories: applications with random node configuration and applications with configurable nodes. With different applications, this paper studies the cooperative decision-making and energy-efficient MAC protocols respectively.In the applications with random node configuration, the network structure is unknown so we focus more on improving network structure, node collaboration and decision-making. Binary decision in traditional detection algorithms will lead to high false alarm rate, and there is "blind spot" coverage problem. Therefore, this paper presents a distributed target detection algorithm based on credit degree - k-CD algorithm. The algorithm: First, forming the virtual node set; Second, matching decision fusion; Finally, using trigger mobile node to solve the "blind spot" coverage problem. Simulation results show that compared with existing algorithms, k-CD algorithm can improve 35% detection accuracy and reduce false alarm rate of 62% in average, and the network life cycle can be extended by an average of 44% in different network coverage. In the applications with configurable nodes, we can configure the network nodes according to practical applications so the network structure and the process of node collaboration are fixed. Energy efficient can be further improving by designing bottom protocol. This paper presents LETD algorithm based on LE-MAC. LE-MAC protocol combines the polling MAC and the scheduling one: part of the sleep-wake schedule can be fixed according to the application environment while maintaining a competitive dynamic MAC mechanism to adjust different network traffic. In LETD algorithm the network is two-dimensional with the false carrier sense and data prediction mechanism. The experimental results show that, LETD algorithm compared with existing algorithms, can make 12% ~ 48% more energy savings.On the basis of the above theory research, we developed a fire detecting system based on wireless sensor networks. This paper describes the modules of the system, especially the design and implementation of MAC protocol and related technical support. Practical test data show that compared with the system with existing MAC protocols, the system can save energy of about 45% and extend the life cycle twice.