| Over the past few years,with the explosive growth of the Internet and easy access to the Web,geodetection information technology developes towards web services.Cloud computing promotes the geodata sharing and interoperability functions towards web services.As business requirements in the field of geosciences become increasingly more complex,individual web services are not sufficient to satisfy these business requirements.Consequently,various distributed individual web services have to be combined to achieve high-level geoprocessing functions.Reusing the web service composition can substantially increase the development speed of geosciences applications-a feat that individual services are unable to achieve.In recent years,research on geodata web service composition has focused mainly on technologies for execution.Also,existing web service composition specifications mainly support direct composition and execution.For execution efficiency,making it possible to discover potential defects during the process design phase to shorten the defect discovery time,reducing the redeployment cost,increasing process reliability and giving a better overall result at a lower cost,geodata service composition model is in urgent need.Also,godata service composition can be proposed for basis of the geodata service composition web application.On the basis of the latest research results related to this dissertation,it presents the geodata service model,geodata service composition model,geodata service compostion model soundness verification,then describes the geodata service composition web application in the design phrase and implementation phrase,finally on the basis of the geodata service composition,the geodata service composition deployment strategy and service composition optimization in the cloud have been discussed.The key contributions of this dissertation include:1.Geodata service composition model has been built based on Petri nets for dealing with asynchronous and concurrent composite service processes.Analysis of the soundness of geodata service composition,e.g.,reachability analysis,boundedness analysis,deadlock analysis and optimization analysis have been developed.In general,procedural programming constructs have four basic patterns: sequential,parallel,loop and conditional.These typical patterns can also be used to describe geodata service composition structure and can be used to form complex patterns.Geodata service composition has been built based on service nets and four basic patterns that deal with asynchronous and concurrent composite geodata service processes.Petri nets and their extensions are also of fundamental interest because they provide modeling approaches for concurrent,parallel distributed systems;define easy graphic support for the representation and understanding of these basic systems;start from state machines to handle the creation and analysis of models;express the main basic concepts in communication,including waiting and synchronization,and being unrelated to a particular implementation language,and provide specifications that are independent of implementation.Furthermore,many validation methods have been developed based on various theoretical results and support tools.It uses Petri nets to validate soundness of the geodata web service composition,and makes it possible to discover potential defects during the geodata web service composition design phase.The geodata service composition produced using this approach can suitably handle unexpected events that may occur at run-time.This can shorten the defect discovery time,which further reduces the redeployment cost,increases process reliability,and gives a better overall result at a lower cost.2.The geodata service composition web application from the design phrase and implementation phrase as a whole has been described.Cutting of the geomodel to achieve visualization and analysis is a very common operation.Here,a geomodel web-based cutting application as an illustrative example of a geodata application that is utilized over the web has been described.In view of the above related work conducted on geodata services,there are two main phases in building a geodata service composition web application: the design phase and the implementation phase.The former phase focuses on the modeling and analysis of the geodata service composition;the latter phase focuses on the implementation of the geodata service composition web application.Service semantics is used for describing geodata service composition to align the technology environment with its business process.Service processes can be modeled as service nets using Petri nets;thus,Petri nets are chosen to model the geodata service composition,and its structural analysis techniques are used to verify deadlock.Finally,the implementation architecture of a geodata service composition web application has been proposed and implemented.This contribution is theoretically and practically relevant because of the advantages offered by service composition web applications for geodata applications,including cost effectiveness,ease of use,flexibility,reusability and ease of deployment.3.The geodata web service composition deployment in the cloud has been discussed.The problem of implementing geodata web application in cloud computing can be defined as determining how to select atomic services to obtain complex services to satisfy the functional requirements and the quality of service requirements.Geodata web service composition deployment in the cloud(i.e.,selecting the optimum required services provided by different service providers in the cloud)is a crucial issue because it affects execution performance.Different services can be distributed over a set of different machines based on service deployment strategies depending on quality of service(QoS)considerations.In general,several QoS aspects may be considered for geodata web service composition deployment,including cost and response time.Therefore,determining the most appropriate strategy is a crucial research problem.A service deployment approach to minimize the inter-service communication time by constructing a service relationship graph(SRG)that describes the interdependence relationships of different services has been develped.According to related studies,the service deployment problem can be mapped to a general graph as a k-partitioning optimization problem.This has prompted us to develop the proposed two-phase approach to solving the graph partitioning optimization problem.A series of experiments have been presented to verify the feasibility of the service deployment approach.The service deployment approach outperforms significantly the modified greedy algorithm which has been used commonly for graph partitioning.4.The network topology optimization in the cloud has been proposed.The problem of the network topology in the cloud can be described how to organize the servers located in different places in the cloud.It is one of the key issues for the effective performance.Considering the cloud resource provider and cloud consumer,the general network topology in the cloud has been abstracted.The network topology optimization model with genetic algorithm and pattern search algorithm to achieve low cost and shortest path has been proposed.The evolution usually starts from a population of randomly generated individuals,and is an iterative process,with the population in each iteration called a generation.In each generation,the fitness of every individual in the population is evaluated;the fitness is usually the value of the objective function in the optimization problem being solved.The more fit individuals are stochastically selected from the current population,and each individual’s genome is modified(recombined and possibly randomly mutated)to form a new generation.The new generation of candidate solutions is then used in the next iteration of the algorithm.Commonly,the algorithm terminates when either a maximum number of generations has been produced,or a satisfactory fitness level has been reached for the population.While pattern search algorithm focuses more on the local optimization.The experiments of the combination of the algorithms can verify the robustness of the network topology in the cloud.5.The geodata service composition optimization algorithm in the cloud has been discussed.Based on the related works,the quality of service parameters can be standardized,optimum service can be selected from service pools,service composition restricitions can be decided,important quality of service parameters can be decided and finally the geodata service composition optimization algorithms can be proposed.The geodata service composition resource provision algorithm has been discussed,the quality of service requriments has been focused on and the optimum resource for the cloud load balancing has been found.Both the resource provision and resource scheduling have been analyzed.The geodata resource provision algorithms in the cloud based on the resources provison parameters have been proposed.The geodata service composition load balancing algorithm in the cloud will be discussed and the parameters such as low response time,high scalability,high resource utilization,high fault tolerance,low computing time and communication time have been focused on. |
