Research On Spatiotemporal Observation Capability Object Field And Computing Service Under The Geospatial Sensor Web Environment

The Geospatial Sensor Web(GSW)is a new type of cyber-physical infrastructure that comprises heterogeneous sensor resources capable of monitoring and collecting various types of geographical environmental information.With the rapid development of the Internet of Things(Io T),sensor,and 5G technologies,the sensor resources in GSW are expected to greatly increase,leading to the situation that sensors are everywhere.The spatiotemporal observation capability is the core of sensor discovery and planning in GSW and is also an important index for evaluating sensor observation efficiency.This thesis highlights that the effective representation and cognition of spatiotemporal observation capability is critical to the efficient discovery and planning of sensors.Given the increasing complexity,diversity,and frequency of disasters and emergency events,GSW mainly promotes the ability of disaster monitoring and information collection by deploying new sensors continuously.However,a limitation exists in understanding the spatiotemporal observation capability at specific geographical locations and times.Questions such as “Which sensors exist at a particular location and time?”,“What are their observation capabilities?” and “How can they be associated and collaborate?” cannot be answered comprehensively and efficiently.This limitation results in two problems: First,although GSW has many sensors,only a few can be discovered and utilized.Second,the observation capability association relations among sensors are unknown.They have hindered the comprehensive and optimized discovery and planning of sensors for observation applications.Geographical information knowledge is a new driving force for achieving the “UN2030 Agenda for Sustainable Development”.Obviously,spatiotemporal observation capability is also a kind of geographical information knowledge that can support the discovery and planning of sensors.Standardizing,openly representing,and sharing such knowledge can help achieve the aforementioned goal.Furthermore,different users typically have different spatiotemporal observation capability cognition demands,thus it is essential to provide an effective knowledge service method for on-demand querying,inference,and calculation on the spatiotemporal observation capability.In general,this study mainly faces three problems.First,there is a lack of integrated mapping between spatiotemporal location and sensor observation capability,which prevents the comprehensive cognition of sensor observation capability on arbitrary spatiotemporal locations.Second,the spatiotemporal observation capability association relations and collaborative efficiency among multiple sensors have not been effectively characterized and calculated,which hinders the comprehensive and optimized collaboration of sensors.Third,the lack of effective spatiotemporal observation capability knowledge representation and open knowledge service mechanism hinders the standard representation,open sharing,and application of spatiotemporal observation capability.To solve these problems,this thesis carries out the following research contents:Firstly,this thesis constructs the Spatiotemporal Sensor Observation Capability Object Field(ST-SOCO-Field)model based on the GIS object-field model to address the problem of spatiotemporal location-based sensor observation capability cognition.The core components of the ST-SOCO-Field are Sensor Observation Capability Particle(SOC-Particle)and Sensor Observation Capability Particle Cluster(SOC-Particle Cluster)with object-field duality,which innovatively constructs the mapping between spatiotemporal location and Sensor Observation Capability Object(SOC-Object).Meanwhile,ST-SOCO-Field utilizes the sequent snapshot model to fully record the spatiotemporal dynamic change of observation capability.ST-SOCO-Field not only supports the discovery and description of SOC-Objects on arbitrary spatiotemporal locations,but also provides a series of cognitive operations to understand spatiotemporal observation capability information like multisensor combinatorial observation capabilities,association relations,and observation applicability.The verification shows that ST-SOCO-Field is superior to the previous field-and object-based modeling methods.It not only has the efficiency of the field model and the integrity of the object model,but also has spatiotemporal dynamic integrity.Overall,ST-SOCO-Field provides an information model basis for the efficient and reliable spatiotemporal location-based discovery and planning of sensors.Secondly,this thesis proposes a Spatiotemporal Observation Capability Association Graph(ST-OCAG)based on graph theory to solve the problem of not knowing observation capability association relations among sensors.ST-OCAG consists of two types of graphs: General Association Graphs(GAGs)and Sensor Observation Capability Clique(SOC-Clique).GAGs mainly calculate and characterize the multi-level and multidimensional association relations between SOC-Objects.SOC-Clique breaks the barrier between binary association relation and N-ary association relation for the first time,thereby answering how arbitrary multiple sensors can be optimally associated and collaborated.Specifically,in order to efficiently solve SOC-Cliques,a “set vector” based approach for identifying a SOC-Clique and a Bron-Kerbosch algorithm-based approach for solving SOC-Cliques are proposed.The verification shows that ST-SOCAG can effectively support characterizing and calculating the observation capability association relations and collaboration efficiency among sensors at arbitrary spatiotemporal locations in ST-SOCO-Field.Compared with other mainstream collaborative planning methods,the proposed method is more efficient and can solve the more comprehensive and optimized collaborative observation plans.Moreover,the proposed method can be further applied to other methods to improve their computational efficiency and quality of results.Overall,by breaking through the bottleneck of not knowing spatiotemporal observation capability associations among sensors,ST-OCAG provides an effective way to support comprehensive and optimized collaborative planning of sensors in ST-SOCO-Field.Thirdly,this thesis proposes a SOCO-Field open knowledge service framework regarding the knowledge representation and service requirements of spatiotemporal observation capability.It mainly includes the ST-SOCO-Field knowledge ontology designed based on the previously proposed models and semantic web technologies and standards,the open knowledge service designed through the extension of SPARQL,and the knowledge service engine designed based on the computation graph model.In particular,the designed knowledge service engine innovatively overcomes the bottleneck of being not able to perform complex computations on semantic web-based knowledge graphs.The verification successfully constructed a knowledge graph based on the designed ST-SOCO-Field knowledge ontology and implemented and utilized the knowledge service engine to conduct a specific spatiotemporal observation capability query and computation service on the constructed knowledge graph,obtaining the required heterogeneous spatiotemporal observation capability knowledge and solving the problem of lacking effective knowledge representation and open service mechanism.In general,the proposed framework provides knowledge and technical support for standard representation,open sharing,and applications of spatiotemporal observation capability.Finally,this thesis designed and implemented the ST-SOCO-Field prototype system based on the aforementioned models and methods,and conducted a flood monitoring experiment based on a heavy rainfall event in Hubei province.The experiment discovered20 heterogeneous sensors in Hubei province and constructed the ST-SOCO-Field with 34SOCO-Fields and their corresponding 34 GAGs included.Furthermore,the ST-SOCOField knowledge graph was generated based on the constructed ST-SOCO-Field.By applying the constructed knowledge graph in the developed system,the efficient and comprehensive computation and representation of observation capability knowledge on arbitrary spatiotemporal locations within Hubei province were achieved,thereby effectively supporting sensor discovery,planning,and on-demand workflow construction under heterogeneous flood monitoring requirements.After further analysis and comparison of the experimental results,the relevant conclusions have shown that the STSOCO-Field and computation and service approaches proposed in this thesis provide a more effective new way of discovering,planning,and utilizing sensors,thus supporting the relevant departments to monitor and responding to disaster events more efficiently and accurately.