Design and implementation of geospatial sensor web node ...

Download PDF
3 downloads 238 Views 482KB Size Report
Comment
professional, with low utility, not easy to use and with a lack of effective validation of model ... paper presents the design and implementation of a sensor web.
Design and Implementation of Geospatial Sensor Web Node Management Prototype System Changjiang Xiao1, Nengcheng Chen1*, Ke Wang1 State Key Lab for Information Engineering in Surveying, Mapping and Remote Sensing Wuhan University 129 Luoyu Road, Wuhan Hubei, China, 430079 *Corresponding author, e-mail: [email protected]

1

Abstract—How to manage the heterogeneous nodes within a sensor web is the key to successful management of a sensor web. Currently, there are masses of problems for node management in other fields, including that they are applicable for mere networks of small scales and can only applied to homogeneous networks, resulting in that they cannot meet the needs of the management of heterogeneous nodes in a sensor web. What’s worse, existing node management systems for a sensor web are extremely professional, with low utility, not easy to use and with a lack of effective validation of model files. To solve these problems, this paper presents the design and implementation of a sensor web node management system (SWNMS) with functions of rapid modeling, information query and edition and visualization of sensor web nodes based on the sensor web node meta-model, referring to the specification of Open Geospatial Consortium Sensor Modeling Language (OGC SesensorML). The extraction of flooding area of Poyang Lake in 2010 is presented as a SWNMS scenario. Keywords-Sensor web node management system; node metamodel; OGC SensorML; node modeling; node query; node visualization

I.

INTRODUCTION

A sensor web [1] is a coherent set of heterogeneous, loosely-coupled, distributed nodes, interconnected by a communications fabric that can collectively behave as a single dynamically adaptive and reconfigurable observing system. The node in a sensor web is an independent entity that performs one or more of the following functions: sensing, computing, storing, directing, and communicating. They interoperate with common standards [2-3] and services [4-5] and can participate in one or more sensor web(s). These nodes can be divided into three catalogues, i.e. sensing nodes, processing nodes, and application nodes, as is depicted in Fig. 1. Sensing nodes refer to various sensor-centric platforms that can directly get property values of observing object(s), including space-borne, air-borne and ground platforms. Processing nodes refer to processing centers that can perform various processing and handling and eventually output all kinds of data products. Application nodes refer to specific services for different applications. Meta-modeling is to build a collection of concepts in a particular area, focusing on restrains and regulations of a model itself and is the further abstraction of a model. Sensor web node meta-model defines the classification systems, attribute

Supported by National Basic Research Program of China (973 Program) (no. 2011CB707101), National High Technology Research and Development Program of China (863 Program) (no. 2013AA01A608) National Nature Science Foundation of China (NSFC) program (no. 41171315), and Program for New Century Excellent Talents in University under Grant NCET-11-0394.

descriptions of sensor web nodes, and additionally provisions the frameworks and guidelines node modeling needs to adhere to. It includes four aspects, i.e. identification, accessibility, state, and performance of a node. The structure of node metamodel is detailed in Fig. 2. The node meta-model facilitates the efficient management of heterogeneous nodes in a sensor web, rapid discovery of distributed resources, unified deployment of heterogeneous resources, comprehensive assessment of node performance and assessment of quality of service through standardized description of various resources and service nodes and the relationships between them within a sensor web using sensor web node meta-model. A sensor web can be thought of as a macro-instrument concept that allows for the spatio-temporal understanding of phenomena which takes place in geographic environments through the coordinated efforts of a large number of nodes of different types [6]. It plays a vitally important role in plan of agriculture, estimation of crop yield, early warning of hazards and emergency response and so on. However, once a sensor web is deployed, the management of such a complex system is a real challenge because of the various heterogeneous nodes in it. These nodes differ in organizations, the ways they deal with problems and communication standards, resulting in their distributivity, heterogeneity, dynamic, virtuality etc. Consequently, how to effectively manage the nodes in a sensor web becomes the key to successful management of the sensor web. While a great progress has been made in research of node management, the specialized study of sensor web node management is still in its beginning stages. Solutions of node management in other fields have some inherent drawbacks, including that some are applicable to mere networks of small scales, that some are in the absent of taking the system having more than one sub-network into account and that some can only apply to homogeneous networks, and consequently cannot meet the needs of sensor web node management. Some existing specialized systems for sensor web node management are extremely professional and with low utility and the model files they produced are in the lack of effective validation. As a result, the systems’ user communities are limited to some professionals with lots of priori knowledge of modeling procedure, and the correctness and validation of model files cannot be granted. To solve these above problems, this paper details the design and implementation of a sensor web node

management system (SWNMS) with functions of rapid modeling, information query and visualization of sensor web nodes based on node metal-model. Among the three functions, rapid modeling of nodes provides two ways for modeling with one based on guide template and the other making use of model tree, suitable for users with little and quantity of priori knowledge of modeling procedure respectively, contributing to a wide range of user communities, having advantages of easiness and convenience for use. The function of node information query and edition provides queries based on node’s type, performance and spatio-temporal location, and supports edition and extension of node information model. The function of node visualization makes it come true to dynamically demonstrate the spatio-temporal locations and attribute information of sensor web nodes based on Google Earth API [7]. The rest of this paper is organized as follows. The following section details the design of SWNMS from perspectives of design principles, architecture, development components and characteristics. Then, in section 3, the implementation of this system is proposed. Section 4 presents SWNMS scenario. Finally, a conclusion is provided and future work is discussed in section 5.

II.

This system is designed to provide a common tool for modeling, querying and visualization of sensor web nodes based on Open Geospatial Consortium Sensor Modeling Language (OGC SensorML) and node metal-model. This paper describes the system in detail from following aspects A. System design principles Sensor web node management system construction is an extremely practical system construction, during which standardization, utility and reliability of system have to be considered. On the other hand, flexibility of system solution is a must to provide extension interfaces for future probable new tasks. Meanwhile, it should be taken into consideration that observation is efficient and advanced, to live up to which advanced design methods and technologies ought to be adopted. In this system, human-computer interaction mode adopted has to adhere to following design principles from perspectives of business and performance. 

Open, advanced and standardized.

Openness shows the live aspect of a system for only when a system is open, it can be enough compatible and ensure former investments being effective still. This can make the system be completed gradually. Technology solutions, including computer systems, operation platforms etc., should be fairly advanced, which can reduce the life cycle of system development and improve system performance. Standardization lays a solid foundation for system construction and meanwhile ensures that a system is compatible with other systems and can be extended. 

Figure 1. Various nodes in a sensor web

SYSTEM DESIGN

Reliable and stable.

Reliability indicates an important aspect of a system’s performance and is determined by robustness and faulttolerance. Stability refers to correctness and robustness aspects of a system. On the one hand, we should ensure that a system function normally for as a long time as possible; on the other hand, the system should have enough robustness which means it can deal with and recover from faults of software or hardware when they occurred unexpectedly. 

Practical and easy to use.

When we say a system is practical, it usually means it can meet the needs of work as much as possible. It should take enough account of convenience and feasibility of data processing in each layer of current business of system user during system construction, namely, user’s requirements of the system shall be the main factor of our consideration. The design of human-machine operation ought to account for human body’s organization and characteristics of our vision. This usually contributes to Graphic User Interface (GUI) which is easy to use and convenient to operate on.

Figure 2. Node meta-model structure

B. System architecture This SWNMS is designed with layered architecture, with three layers in total, i.e. resource layer, business layer and presentation layer, as is shown in Fig. 3.

Resource layer provides all kinds of sensor web node resources, including sensing node resources, processing node resources and application node resources. Take sensing node resources for example, they contain various sensor resources, for example, space-borne (meteorological satellites, atmospheric satellites, cartographic satellites and mitigation satellites etc.) sensors, air-borne (uninhabited aerial vehicles, air-ships and balloons etc.) sensors and ground sensors (thermometers, hygrometers, photometers and radiometers etc.), all of which are stored in model files in eXtensible Markup Language (XML) [8] on local disks. These model files all adhere to the specifications of OGC SensorML [9] and predefined XML Schema. Node resources in resource layer can be created, edited, queried and visualized through this system, providing data for upper layer and thus lay a foundation for the whole system. Business layer is the core of SWNMS, which defines a serial of businesses and operations of the system, including node modeling, node information query and edition visually and node visualization dynamically. Presentation layer provides a serial of GUIs through which system users and the SWNMS can interoperate with each other. This system presentation layer is developed within .NET Framework3.5 [10], through those GUIs of which users can accomplish businesses and operations defined by business layer. Take sensor web node resource modeling for instance, users can select corresponding templates for modeling through user interfaces, type in values to the right places on the modeling interfaces to establish a model instance rapidly. Interfaces of this layer are designed to be simple and easy to use as much as possible and great efforts have been made to make interfaces under the control of system users but not the opposite, to reduce users’ burden of memory and to keep consistency among different user interfaces [11].

C. System development components Development components of this system mainly include .NET Framework 3.5, Lucene.Net [12] development kit, and the three-dimensional visualization platform, Google Earth, as is depicted in Fig. 3. They all play vitally important roles within the whole system. Among these development components, .NET Framework3.5 provides the whole common visual development framework, and of course is the core development component of the system. Lucene.Net is introduced into C#.NET as a three-party development kit, which is used to establish indices of locally stored node model files and execute various queries among them. Google Earth, a three-dimensional visualization platform for geographical information [13], can load spatio-temporal information of sensor web nodes and demonstrate them dynamically. D. System Characteristics This sensor web node management system adopts MVC design pattern [14], making models, views and controllers apart from each other, which can decrease coupling between function modules and at the same time, increase aggregation within each function module, consequently making it easier for modular design of the system. Meanwhile, the system development is object oriented (OO). When a model is created following a template, the template is thought of as a class; the model is seen as an instance of that class; the process of modeling is regarded as a process of instantiation of a class; operations on model files can still be seen as operations on instances of a class (classes). As a result, the flexibility of the system and reusability of codes are increasingly improved. Besides, a lot of interfaces remain for further extension of functions as needed, which improves the extensibility of the system. The system is implemented based on .NET Framework3.5 development platform, by making use of Visual Studio 2008 an integrated development environment, through the introduction of Lucene.Net development kit and invocation of Google Earth APIs and using C# program development language. III.

Figure 3. Architecture of SWNMS

SYSTEM IMPLEMENTATION

A. Visual modeling of node resources Visual modeling of node resources makes it possible for users to model sensor web node resources rapidly based on node meta-model. It provides two ways for users modeling according to their profession, namely modeling with templates and classification tree. The process of modeling with these two methods is depicted in Fig. 4.The former one takes the form of Question and Answer (Q&A) to guide users to select the right template for modeling. Then users type in suitable and correct values to the corresponding places on the template according to their priori knowledge of node resources and requirements of processing or define special parameters needed to establish a model meeting their satisfactions rapidly. This is really a convenient and wise way of modeling suitable for users without too much priori knowledge of structures of SensorML. The latter one makes full use of classification tree of sensor web nodes, on which users can select the right type of nodes needed and create a model instance rapidly through operations provided by context menus bound to each node including Add

a node, Delete a node and Edit a node and is suitable for users with much priori knowledge of structures of SensorML and node meta-model. Providing these two means of modeling, the system can cover as many as users ranging from fresh hands to professionals, and the flexibility of the system is improved to some degree in consequence.

Figure 5. The interface of node model edition in SWNMS

Figure 4. The process of modeling with two ways in SWNMS

B. Visual edtion of node resources Visual edition of node resources enables the edition of node information models already established formerly. Once the node information model has to be updated, users can utilize this function module to modify some parts of the corresponding model files purposely.

C. Visual query of node resources Visual query of node resources supports various queries of node information including queries based on type, performance and spatio-temporal location of a node (nodes). Queries based on node type can help users get nodes of the specified type. Queries based on node performance have twofold meanings. On the one hand, a user can query the performance set of the specified node. On the other hand, a user can specify a performance set as a query filter to get the set of nodes meeting these conditions, which really helps a lot in selection of nodes with required performances in the process of a specified observation mission. Queries based on spatio-temporal location of a node (nodes) specify a restrain of a time period and a spatial boundary (a rectangle on the surface of a sphere of the earth) as a query filter to get the set of nodes whose observation scope can completely or partly cover the specified extent during the specified time period, and provide topological relationships between the specified extent and the observation scope of each output node. Consequently, selection of nodes whose observation meets the specified spatio-temporal requirements comes true.

The information model can be loaded into the system through a serial of operations and will be displayed in a tree structure on left part of the main user interface. When a user clicks on a node, all its sub-nodes’ information will be shown in the form of table on the right of the main user interface with each table cell editable. Then the user can click on the right cell and edits its value as needed to get the node information model updated to newest, as is shown in Fig. 5.

Figure 6. The interface of node model visualization in SWNMS

D. Visual and dynamic demonstration of node resources Visual and dynamic demonstration of node resources mainly demonstrates information of nodes’ spatial location and attributes. The system loads these nodes’ spatial and attribute information to Google Earth through Google Earth APIs. This information is obtained in near real time according to the updating frequency specified by users, and will be updated on Google Earth instantly, as is depicted in Fig. 6. IV.

SWNMS SCENARIO

Floods [15] are one of the natural disasters happening most frequently in China, especially in the middle reaches of Yangtze River, e.g. Poyang Lake. Every year floods destroy parts of the environment and man-made infrastructures and threaten human lives, thus there are certain requirements for flood crisis management, e.g. the extraction of the flooding area [16]. Take the flood in Poyang Lake in 2010 as an example, to achieve the goal of extraction of this hazard, three types of nodes are needed, i.e. sensing node (e.g. MODIS, to acquire remote sensing imageries of the flooding area), processing node (e.g. data processing center of MODIS, to extract and calculate flooding area through data processing of remote sensing imageries of MODIS), and application node (flood hazard evaluation center, to evaluate the loss of population, property etc. quantitatively according to the calculated flooding area and guide post-flood reconstruction),. Utilizing the SMNMS, these above types of nodes can be modeled, queried based on node types and node performances, and visualized on Google Earth by loading the spatial information and attribute information of nodes to it. V.

from enough to support selection of nodes meeting the requirements of a specified observation mission. Hence, more advanced queries such as queries based on accessibility and availability of nodes have to be developed. In addition, only the attribute information of nodes, but not the relationships between nodes which are necessary for effective management of sensor web nodes, has been taken into consideration. This can be another aspect of our future work.

CONCLUSIONS AND FUTURE WORK

This paper has designed and implemented a sensor web node management system based on node meta-model with enough investigation of currently existing node management systems. The system has three layers, i.e. resource layer, business layer and presentation layer and mainly four functions, i.e. modeling, edition, query and visualization of node resources which are defined in business layer of SWNMS. This system provides a uniform way for sensor web node management by rapidly modeling, efficiently querying and near real-time visualization of heterogeneous nodes within the context of a sensor web. Furthermore, it lays a solid foundation for establishment of nodes’ in and out mechanism and of sensor web’s architecture design and transform pattern. It overcomes drawbacks of traditional systems, including that they are too professional, in the absent of effective validation of model files and with low utility. Providing various guide templates for modeling, hiding XML codes from users and validating model files etc., it is convenient and easy to use. Only several templates have been established thus not all kinds of nodes’ modeling are supported. We consider optimizing our current templates to improve its commonality as well as increasing templates to support modeling of more kinds of nodes in our future work. Besides, as with queries, only some basic queries based on type, performance and spatiotemporal location of nodes have been realized, which is far

REFERENCES [1]

[2]

[3] [4]

[5] [6]

[7] [8] [9]

[10] [11] [12] [13]

[14]

[15]

[16]

NASA, Report from the earth science technology office (ESTO) advanced information systems technology (AIST) Sensor Web meeting, 13-14 April 2008, Orlando, FL. GreenBelt, MD: NASA Earth Science Technology Office. N. Chen and C. Hu, “A sharable and interoperable model for atmospheric satellite sensors and observations,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. vol. 5(5), 2012, pp. 1519-1530. M.. Botts, “OpenGIS® Sensor Model Language (SensorML) Implementation Specification,” OGC® 07-000, ed: Open Geospatial Consortium, 2007, p. 180. N. Chen, L. Di, G. Yu, and M. Min, “A flexible geospatial Sensor Observation Service for diverse sensor data based on Web services,” ISPRS Journal of Photogrammetry and Remote Sensing, vol. 64(1), 2009, pp. 274 –282. Z. Chen, N. Chen, L. Di, and C. Yang, “A flexible Data and Sensor Planning Service for virtual sensors based on Web Service,” IEEE Sensors Journal. vol. 11(6), 2011, pp. 1429-1439. M. Mekni, “Intelligent Sensor Web Management Using Multi-Agent Geo-Simulation Techniques,” Proceedings of the International Conference on Applied Computer Science, p. 540, Iwate Prefectural University, Japan, 4-6 Oct., 2010. Google, Google Earth API, https://developers.google.com/earth/?hl=en, last accessed on 22nd Feb., 2013. World Wide Web Consortium (W3C), XML TECHNOLOGY, http://www.w3.org/standards/xml, last accessed on 22nd Feb., 2013. N. Chen, C. Hu, Y. Chen, C. Wang, and J. Gong, “Using SensorML to construct a geoprocessing e-Science workflow model under a sensor web environment,” Computers & Geosciences. vol. 47(2012), 2012, pp. 119-129. Microsoft, .NET Framework3.5, http://msdn.microsoft.com/en-us/library /vstudio /w0x726c2(v=vs.90).aspx, last accessed 22nd Feb., 2013. J. Tidwell, Designing Interfaces, 2nd ed., America: O’REILLY®,6 Jun., 2011. O. Gospodnetic, and E. Hatcher, Lucene In Action. America: Manning Publications Co., 2005. A. Chen, G. Leptoukh, S. Kempler, and L. Di, “Visualization of NASA Earth Science Data in Google Earth,” Proc. SPIE7143, Geoinformatics 2008 and Joint Conference on GIS and Built Environment: GeoSimulation and Virtual GIS Environment, 714329 (November 03, 2008). IBM, Model-View-Controller design pattern, http://pic.dhe.ibm.com/ infocenter/wchelp/v6r0m0/index.jsp?topic=%2Fcom.ibm.commerce.dev eloper.doc%2Fconcepts%2Fcsdmvcdespat.htm, last accessed on 22nd Feb., 2013. Kridsakron, A., Chen, N., Peng, C., Yang, C., Gong, J., 2012. “Flood detection and mapping of the Thailand Central plain using RADARSAT and MODIS under a sensor web environment,” International Journal of Applied Earth Observation and GeoInformation. vol. 14(1):245-255. S. M. El-Sayed, “Flood Management using remote sensing,” avaliable online at http://www.unesco-ihe.org/content/download/2113/22004/ file/Flood%20Management%20using%20remote%20sensing.pdf.