Solutions For Experimental Data Service System

1. QuestionsThe Tectonophysics Unit of State Key Laboratory of Earthquake Dynamics has ahistory longer than 20 years. With development of geosciences and experimentseismology, the observation content of experiments becomes abundant constantly.Meanwhile, development of electron and computer technology makes the updatecycle of data acquisition systems in the laboratory shorten at full speed, and raises thedata acquisition resolution and speed by a large margin. These progresses have offeredmore abundant and more reliable experimental data for scientific research.Through putting the data in recent years in order and making classification, it isclear that the experimental data have some characters, such as great amount of data,complicated and various data structure, and data diversity in style. This has brought agreat deal of inconvenience to data analysis and data processing, mainly showing inthe following aspects: It is a challenge to store and backup data reasonably as the quantity ofexperimental data is increasing greatly. Data is lost easily by the intricatenonautomatic mode of data management. Users acquire experimental data by downloading from network or copying viaCD. It aggravates the burden of network and occupies a lot of user's storagespace. Data files with different formats are generated by different data acquisitionsystems and there is no standard for reference. Users have to know the fileformats clearly, making data access and processing more and more difficult. In daily scientific research, the methods of data processing and analysis areflexible and changeable. However, some of common methods have not beenaccumulated and shared reasonably. As a result of writing and compiling samedata processing codes repeatedly, data processing time is increased.2. Goals The existing data processing mode can not meet the complicated and specializeddemand in the laboratory. To the characters of experimental data and data processingprocedures, a platform should be established to reduce the shortcoming existing in thework mode at present and to improve working efficiency. The platform should bebased on network. It is able to implement automatic management to experimental dataand offer data with an unified format and the unity data access interface to users torealize data share. The system should not only support accumulation and sharing ofresources of the data, but also support accumulation and sharing of data processingalgorithms, which are based on network. It can offer data calculation environments tovarious kinds of users. Generally, users can process data without need to download agreat deal of data to local computer at first. What the network transmit are onlycalculation orders send out by users and calculation results returned by the server.This study is aimed to establish the system of data management and data serviceexploringly in the laboratory. Firstly, it puts forward the corresponding solution to theshortcoming in the work pattern at present. Then it analyses and chooses relevanttechnology. Finally it takes acoustic emission (AE) data as the example to examinethe theory of the data service system. It offers the theoretical direction and technicalsupport for setting up and constructing experimental data service systems intectonophysics laboratory.3. Solution SchemesIn order to solve various kinds of problems in experimental data managementand using and constructing the data service platform of the laboratory, Thecorresponding solution schemes are proposed as described below.1) Data ManagementFirst of all, standardizing experimental data use SQL Server database and makingit become legible two-dimensional tables, which reduce the structure complexity ofdata.Secondly, programming the Universal Experimental Data Access Interfaces tomake programmers be able to obtain the content of sections of each word in theexperimental database but needn't to understand the structure of the database and themethod to visit it. The interfaces reduce the degree of difficulty of data visit greatly.Moreover, according to the data management procedure of the laboratory,programming corresponding software to realize automatic data classification and saveit on units of the experiment in the procedure of copying data from data acquisitionsystems to the data server.Finally, programming software runs automatically in the server and writesexperimental data saved in the server into database tables regularly.2) Utility of DataAccording to special demands, end users of experimental data can be dividedinto three kinds: common scientific researchers, professional scientific researchersand programmers. Client program design includes two modes: B/S (Browser/Server)pattern and C/S (Client/Server) pattern.A common scientific researcher is the user who browses data and does somesimple processing. B/S (Browser/Server) pattern is chosen to design client softwarebecause it is much easier to use and more commonable. If users setup an InternetExplorer, such as IE and Netscape, in local computer, they can access and deal withremote experimental data via web pages. The web server sends calculation results tothe user.A professional scientific researcher means that who needs to do complicated dataprocessing frequently, such as scientific researcher at a laboratory. Web pages areunable to offer stronger functions to this kind of users. Software designed via C/Spattern can remedy the shortage of web pages. It is able to provide quicker dataprocessing speed and friendly interface to users. But users of the software must installit in local computer.A programmer is the scientific researcher who has the ability of programming.His responsibility is to write code and process data. He can realize new algorithms byprogram languages according to the need of scientific research. The system provideshim accumulated and shared data processing methods through network.3) Network Environment of LaboratoryUpdating servers' hardware and rebuilding network are means to reducebottle-neck of the system.4. Technology ApproachesExperimental data service systems are set up on the foundation of thedevelopment of computer technology. The progress of computer technology promotesdata service system's realization. Choosing one or several kinds of suitabletechnology to solve the problems that are faced now is a key step. Based on theschemes presented above, some popular computer technology are analyzed andcompared in the thesis in order to choose the suitable technologies to apply on thesystem construction.1) Accessing DataODBC(Open Database Connectivity), OLE DB(Microsoft Object Linking andEmbedding Database)and ADO(ActiveX Data Object)are popular technologies ofopen data accessing. Generally, they are used in various program platform. Althoughin the Microsoft Visual Studio .NET program platform it can be run normally,ADO.NET is the best choice in the .NET Framework because it can provide data forusers offline and integrate XML very well. Based on the program developing platformand experimental database, a suit of Universal Experimental Data Access Interfacesare established. The interface screens the database structure and the professionalmethods of database accessing. Programmers acquire all of data in the database easilyby means of invoking methods of the interface.2) Accumulation and Share of Data Processing MethodsNow, accumulation of data processing methods relies on personal software torealize. Most methods are integrated in the software that individuals wrote. It is verydifficult to draw to reuse for the new application program. DLL, ActiveX and DCOM(Distribute Component Object Model), can realize the accumulation and share ofdata processing methods. But it has certain degree of difficulty and the commonabilityis relatively poor. The methods of data processing based on network using WebService will be more commonable and easier to develop. No matter which kind ofdeveloping platform is used, it can all get and use interface of the data processingalgorithms through network. Users needn't understand how the algorithms arerealized.A program developer can envelop data process methods as Web Services andrelease at network very easily. These shared methods will be used by other programdevelopers.Web Service is based on XML, which obtains the best compatibility andsimplifies development. But it causes losses on performance. The .NET Remotingtechnology is the other solution of the system: to use TCP channel and the binaryformat to process and transmit data. It can reach the best performance of distributeddata processed by this way. The accumulation and share can be realized by meansof .NET Remoting too. The technology is able to integrate the new data processingmethod on the server and invoked by remote client software. But the realization relieson the .NET Framework platform, that is to say, all users' computers must installthe .NET Framework in order to support it. So there is a certain restriction on thechoice of the developing platform in this technology.3) Installation and Maintenance of Client SoftwareThe software adopts the B/S pattern only needing to setup and maintain at theend of the server. The software in the server is updated and the web page content atclient is renewed.The installation and update of the client software which adopts the C/S pattern ismore difficult. Users download or copy setup files to local computers and install it.This is the common way to install software now. But the users can not obtain thenewest version of the software in time by this way. The dynamic update in SmartClient technology is able to update the software in client automatically. It means thatusers download the windows installer from the server and run it on local computersfor the first time. If the software in the server is updated, the client software will bedownloaded and updated itself while the user runs it.5. Laboratory Data Service System for Acoustic EmissionBased on the work mentioned above, Laboratory Data Service System forAcoustic Emission is set up tentatively. First of all, according to the characteristics ofAE data, AE data tables' structure is created. The system realizes uploading data filesfrom AE data acquisition systems to the data server and set up the preliminaryexperimental data management rule. The rule of experimental data access interface isformed and the Universal AE Data Access Interface which faces the database isestablished. The system realizes several kinds of data processing methods on thisbasis. The data access interface and the methods are modified and enveloped as WebServices and pass the tests.On the basis of the aforesaid, two sets of data process application programsbased on network are realized.6. Innovation of the SystemBased on the study of the goals, the framework and technologies of theExperimental Data Service System, the AE Data Service System is established toprove the solutions. It can be concluded that the Experimental Data Service Systemovercomes shortcoming in the working mode at present. It improves the datamanagement and use efficiency. The innovation mainly shows in the followingaspects:1) It unifies the raw experimental data files with database and realizes thestandardization of the experimental data. Software developer does not need to dealwith various kinds of data files. The efficiency of data share and access is raised.2) Most of gathering and management of the experimental data are finishedautomatically by the computer. It reduces manual intervention and improves theefficiency of management and reliability of data.3) The Universal Experimental Data Access Interface and the accumulation andshare of the data processing methods on network reduce the difficulty of softwaredevelopment, enable more researchers who have ability of programming to participatein the data processing, improve the rate of utilization of the data and promote thealgorithms for the experimental data.4) The system not only realizes the sharing of data but also realizes the sharing ofcalculation environment. The data processing is fulfilled on the server. No mattercommon scientific researchers or professional scientific researchers, the friendlyclient software may be offered through the system. Users send out the calculationcommand to the server and receive the results from the server. It guarantees theprocedure of data processing unobstructedly and fast. As more and more scientificresearchers can participate in the experiment data analysis, more research results willbe generated and the work efficiency will be enhanced.7. Problems and ExpectationIt is feasible to manage and use the experimental data automatically andconveniently based on the Experimental Data Service System. It can improve thepresent work mode and enhance the work efficiency. However, the system is at theexploratory developing stage now. So there are some problems, such as that thefunction is not perfect enough. All of these problems will be solved progressively inthe future. But it still provides the theoretical foundation and technical support for theoverall construction of the system.Because the introduction of the new system will change the existing work mode,users need time to understand, to study and to accept the new system. There is a longway to the users of the system to adapt to the new work mode.Experimental Data Service System has a good prospect of application. Itssetting-up will make the management and utilization of the experimental data presentsa new situation in the future.