Research On Fault-Tolerant Event Boundary Detection Algorithm For Wireless Sensor Networks

Wireless Sensor Networks (WSNs) is a hot research field which has got much attention internationally in the21st century, it has been widely applied to many aspect in our society. As a base of the WSNs application, event boundary detection is an essential research subject. WSNs is different from the traditional network, it is a type of self-organized network formed by many cheap sensor nodes through wireless communication. Because of the feature of the WSNs, the sensor node is energy constrained, and the invalid and erroneous node localization can easily lead to node fail. So the event boundary detection algorithm must be energy efficiency and faulty tolerant. In this paper, we have proposed two kinds of distributed faulty tolerant event boundary detection algorithms. Both of the algorithms have the advantage of consume little energy, strong faulty tolerance and high detection precision.The first algorithm is Normal Vector based Fault-tolerant Event Boundary Detection Algorithm. In this algorithm, we map the detection value to3D surface, the3D coordinate was composed of the latitude, longitude and the detection value of the attribute. It is impossible to detect all location of the detection area, so we regard the sensor nodes as a sample of the area, and all of the nodes consist of a mesh surface. The angle of the normal vector between the surface and the horizontal plane is big in the event boundary, but small in both event area and non-event area. This is mainly because the detection values between event area and non-event area have sharp distinction, but the difference in event area or non-event area is subtle. Based on this feature, we compare the angle of the normal vector between the mesh surface and the horizontal with a predefined threshold to determine whether the node is an event boundary node or faulty. The result of the simulation shows this algorithm has a good performance.The second algorithm is Entropy based Single Node Trigger Faulty-tolerant Event Boundary Detection Algorithm. All of the nodes must exchange messages with its neighbors to determine whether the node is an event boundary node if the algorithm is based on spatial correlation, and the energy of sensor nodes mainly is consumed in transmitting data. In WSNs, the node not only has the spatial correlation but also has the temporal correlation, so this algorithm is based on it. The entropy is considered to determine whether the area of the node located in has an event or not. If the node detect an event, it will exchange message with its neighbors to check the result. This algorithm don’t need to transmit data until it detect an event, so the energy costs of the transmit are substantially lowered, and prolong the lifetime of wireless sensor networks.In order to verify the performance of the proposed algorithm in this paper, we used Matlab to complete the data simulation. We build the network and then deploy the sensor node with faulty probability p. The algorithm was executed in each node. The result of the simulation shows that our algorithms has high performance.