The Research On Earth Observation Sensor Information Resource Modeling And Management

Sensors play a fundamental role in our daily activities, while play an important role in environmental monitoring, intelligent information management, emergency management and security defense system. But the technology is based on the heterogeneous networks of heterogeneous sensors. These sensors are difficult to be discovered and accessed,and the process of their observations usually been limited to isolated systems. Even if the interoperability of a sensor system can be achieved to some extent, it costs a lot of money, and the system is difficult to maintain and extend. Few sensor systems have the capabilities to find, access and process sensors accurately. Most current sensor application systems are based on specific sensors and sensor networks, and are rely on equipment, transport protocol and hardware environment, it's difficult to find, access and process interdisciplinary sensors.Along with the continuous development of sensor technology and sensor applications, it needs a common framework to describe the various types of sensors and process systems as well as their interlinkages urgently. Sensor Model Language is a information model of OGC (Open Geospatial Consortium) Sensor Web Enablement(SWE) framework, it can provide the information which needed by sensor access, observation position, and data process.In SensorML, sensor physical platform and logical algorithm can be described as process or process chain,and be implemented in the public workflow engine. SensorML standard focuses on the functional model, not the hardware description model. The method provides a strict model based on parameters and be independent of platforms, while as a mathematical model, it's able to create mapping relations between platforms and process goals.Process is the basic element of SensorML, and various process are compounded to process chain, this can use sensor networks resources effectively. From this perspective, SensorML process chain model can be seen as a process description language unrelated to specific implementation. SensorML defines an open framework that can support a variety of data standards and service standards. It can meet different application needs by restructuring the existing sensor resources to process chain according to demand. SensorML provides information model for SPS (Sensor Planning Service), SOS (Sensor Observations Service) and SAS (Sensor Alert Service) of SWE sensor services.This paper analyzes the basic framework for sensor information sharing firstly, and analyzes sensor modeling method based on the SensorML standard to propose process-oriented modeling approach. It also designs the sensor process chain driven engine based on sensor process chain. The main research activities of this paper include the following:1, sensor information sharing model researchThe sensor information sharing mentioned here not just the sharing of sensor physical hardwares, but further sensor observation data sharing and the sharing of data process methods. This paper analyzes the characteristics of sensors, sensor networks, sensor observation objects and sensor observation results based on the study of sensor information sharing framework. It also analyzes the connotation of sensor information resource sharing framework and content-level sensor information resource sharing framework, as well as the concept of sensor information resource sharing model.This is the basis for the study of this paper.2, sensor modeling method based on SensorML researchSensorML provides a general framework for sensor modeling, it's an abstract sensor modeling specification. Based on the analysis of SensorML,this paper establishes the sensor modeling templates according to the characteristics of different sensors.It establishes the remote sensing satellite sensor modeling templates based on ISO 19130,and defines services process templates and data services templates according to OGC's Web services. Based on the templates it uses process-oriented modeling approach. Template is the smallest unit of SensorML modeling, process-oriented sensor modeling approach is proposed according to the template, and this method uses a recursive evolutionary way to build sensor models.3,the design of SensorML process chain driven engineThe key concepts of SensorML are sensor process and process chain model. And this model is independent with specific type of implementation. SensorML uses digraph approach to describe the relationship between process and constructs a process chain.Treating SensorML process chain as a mean to integrate sensor information resources can separate the specific sensor platforms, sensor data models, and sensor process methods, and achieve the minimum coupling between the levels of sensor resources. SensorML has nothing to do with the specific implementation based on the XML encoding. Based on the analysis of the process chain,this paper proposes SensorML abstract process chain and the process chain example,also designs and implementes SensorML sensor process chain driven engine including process chain example, process chain validation and process chain transfer.4, process chain driven case studiesAt last,this paper establishes a Terra satellite sensor model, and designs the NDVI calculation of non-physical process chain model based on that sensor model. Also it instantiates the MODIS abstract process chain model for MODIS 1B data and the border of Hubei Province data,and runns this process chain example in the process chain driven engine, it's output is NDVI images of Hubei Province. This paper describes instantiating and transferring of NDVI process chain in detail based on NDVI non-physical process chain model, and validates the rationality of sensor information resource sharing model based SensorML and the feasibility of process chain driven engine.The main innovation in this paper can be summarized as the following:1, This paper begins to analyze the basic functions of sensor from the sensor as an extension of human sense,then defines the content of sensor information resource sharing by definition of basic functions. It summarizes the concept level of sensor sharing, including five major parts such as ensor hardware, sensor networks, observing phenomena, observation results and sensor process methods.2,This paper proposes sensor information sharing model framework based on sensor resource sharing framework, including the basic models of the static information of observation objects, observation results, sensor space-time location and sensor metadata.It also establishes sensor process information model based on the static information for SensorML process chain model,which can fully integrates with other types of sensor information model standards.3,This paper creates different sensor application templates based on different sensor standards respectively using the templates as the smallest unit for the modeling,and uses SensorML process chain as the core according to sensor information sharing model.It determines the meaning of sensor resource management which uses sensor model as the basic elements,and establishes resource management of sensor resource index and remodeling modeled by SensorML.4, This paper designs sensor process chain driven engine based on SensorML process chain model to achieve four basic functions such as the process chain analysis, process chain validation, handling routing and handling of actuator modules.5, This paper sets up SensorML model and MODIS NDVI abstract process chain model of Terra satellite according to Terra MODIS information.Based on this study,the mainly focuses required in further research as the following:1,This paper uses basic vocabulary matching without considering semantic background in the abstract instance of process chain. In the future work, it needs to introduce semantic knowledge, and achieve instance of the process chain based on semantic implementation,which reduces the human-computer interaction in instance of the process chain.2, This study is based on SensorML sensor modeling standard under OGC's SWE framework, but the standard has no description of process control,so we consider the introduction of the BPEL or XPDL process description standards to SensorML standard model in the next step. This study expands the description of process by using standardized process control in SensorML,or transfers standardized process model nestedly in SensorML process chain and makes SensorML a sub part of standardized process model, and improves the interaction and interoperability between sensor process chain and the common standard process model.3, The sensor classification criteria used in this paper is the four types sensor classification of SWE, however, the standard definition of these four types sensor is too abstract to finding and positioning sensors. In later studies we can focuse on the sensor further classification, using a variety of classification index to classify sensors.4, The process chain driven engine of this paper is designed in a network environment,without considering the distributed parallel process of the calculation and the asynchronous process. Full-text describes the process chain model by a directed acyclic graph, you can use Petri nets to organize all distributed sensor resources to meet the long aging process of scientific computing.