Understanding the impact of planarized proximity graphs on toxic gas boundary area detection

Detecting the diffusion boundary of toxic gas is an important research issue in petrochemical plants. There have been many research efforts on this issue. However, detecting an accurate absolute boundary is still less likely to be achieved, because of the extreme environmental sensitivity and easy diffusion characteristic of the toxic gas. Therefore, it is more practical to detect a boundary area instead of an absolute boundary. In this paper, we analyze the impact of four planarized proximity graphs on the accuracy of detecting the boundary area, based on a boundary detection algorithm. By extensive experiments and analyses, we obtain three important observations to understand the different performance of four graphs on different detection scenarios (e.g., different numbers of sensor nodes and different radii of toxic gas leakage). Moreover, we learn the factors that influence the accuracy of boundary area detection. The learning results can be used to direct the strategy design of boundary area detection.