Research On Event-driven Method For Dynamically Observing Forest Fire Under Sensor Web Environment

The fire is one of major disaters which are the most common and general threats to public safety and social development. Each year, a large number of forest fires occur all over ther world. They cause serious damage to the ecological environment and huge losses to people’s lives and property. How to conduct a comprehensive observation to forest and timely access to environmental information for forecasting the occurrence of forest fires in order to minimize the losses caused by them has been an important research topic of human being’s concern. Traditional observation use manual and semi-automated observation methods, which is cumbersome and inefficient and thus is not conducive to the timely discovery of fire dangers. With the development of Earth observation technology, more and more ground, air and space sensors has joined the ranks of environmental observations, which enable human being to faster obtain the forest environment information. Since the end of the last century, the rapid development and integration of computer, communications, microelectronics and other domains gave the birth to the sensor web technology. This technology integrated resources needed by environmental observation, including perception, storage, communication, network, compution and other resource. It can gratly improve the level of automation to discovery, access and process environmental information and thus improve the efficiency of environmental observation. In the past ten years, sensor web technology has gained rapid development and has been studied and applied in many fields. However, how to apply sensor web technology in forest environmental observations to improve the level of forest fire observation is still to be in-depth study.In the field of earth science, most of researches and applications on sensor web technologies are based on series of OGC specifications. In order to promote the openness and interoperability of geographic information, OGC has developed servral series of service specification, including OGC web service specifications and sensor web service specifications. These specifications all design for service-oriented architecture (SOA), which can integrate existing resources very well so that they work together. However, the most suitable mode of SOA is request-response, which is the lack of reactivity. Environmental observation is a perception process, in which active reaction to environmental changes is also very important except that passively access to environmental information. To make up for this deficiency, OGC try to apply event-driven approach in the service specifications. After10years of efforts, the existing research achievements have been able to support some event-driven application in observations. As present, related studies is still less at home and abroad. This paper aims at the needs of the dynamic observation of forest fires and studied some key technologies required for event-driven application of forest fire observation in sensor web environment. The main work and innovation points include the following aspects:(1) For the requirement of event-driven observation method of forest fire, the paper reviewed and summarized the basic principles and methods of forest fire observation, the research and development of service specifications related to sensor web applications, event-driven technology, event-driven technology in sensor web, and pointed out the basic idea of event-driven applications in sensor web environment.(2) The methods for hierarchical division, classification and modeling of forest fire observation events. They are the basis of the design of event-driven applications. There are a variety of levels in events, but there are no clear hierarchical distinctions in events that people talk about in their daily lives. This distinction is necessary in event processing because the handling methods of different level events may not be the same. The paper analysed hierarchical relationship between observations and environments, and gave three abstract leves of observation events. Based on a forest fire prediction model, the events in the process of forest fire observation was studied and classified. In the end, a conceptual model of observation events was proposed in the basis of the above studies.(3) The processing methods of forest fire observation events. The event handling is the key of the implementation of event-driven application. The paper mainly studied sevel key technologies of the handling of observation events, including the organization of observation events, the filtering of observation events, the handling of complex observation events and the modeling of the processing flow of observation events. The support for these key technologies in sensor web service specifications was also analysed. Then, the implementation methods of the first three techniques were studied on the basis of a forest fire predictin model. Finally, the finite state machine was used to model processing flow of observation events.(4) Event-driven architecture applied to forest fire observations. Event-driven architecture is the basis of the design and implementation of event-driven applications. They generally consist of four prats:event generation, event channel, event processing and event consumption. The paper focuses on the requirements of forest fire observations to analyse implementation components in four parts, and then built and analysed the processing flow of observation events in detail on the basis of the components. Next, a conceptual framework was proposed on the basis of the flow and the existing data service infrastructures. In the end, based on the framework, an event-driven framework applied to forest fire observations was designed with sensor web technologies.(5) Encoding observation events and their modes. Encoding observation information using O&M (Observations and Measurements) specification and encoding event patterns using EML (Event Pattern Markup Language) specification are the research hotspot and vision of sensor web technologies. The paper encoded observation events using O&M specification and encoded their models using EML specification, and matched them well in the filtering of observation events.(6) Based on a forest fire predition application, a simulating testbed of forest fire observations was established to verify the feasibility of the models and methods proposed by the paper. The experiment was divided into four phases according to the event-driven architecture. Several key technologies were tested in each phase. Additionally, the paper illustrated the running of the event-driven architecture using the sequence diagram of the experimental task.