Research On Real-time Search Of Entities For Internet Of Things

With the continuous development of information technology, Internet of Thingshas got much attention of community.It was widely used in industry, agriculture,smart home,express logistics,equipment monitoring and other fields. However, a lotof entities are contained in Internet of Things.Leading to a mass of real-time dataincluded in it. How to process and analyze the mass of entities? How to search foruseful information from the returned real-time data? How to predict the unknownfuture state through information? These all have become the focus of the study inInternet of Things.Here conducted a research aiming at searching for entities in Internet of Thingsand realized real-time search for entities.First, in the physical structure and dataacquisition.Tools of MetaSeeker and Watir and Nokogiri were used.Real-timecollecting information based on MetaSeeker used real-time page monitoringmechanisms.It begins real-time grasping as long as the page information changes.Ittakes up little system resources. Real-time collecting information based on Watir andNokogiri mainly used Watir to load the dynamic pages of Internet of Things.Thenused Nokogiri to analyze the HTML and get the document in the HTML of thepages.Found the node of needed content in the pages though CSS Selector. Get thecontent at the moment to realize grasping the real-time information on dynamic webpages of Internet of Things at a fixed period and get the perceived information ofentities.For the search in Internet of Things, search engine Lucene was used. Lucenedoes parse to entity data.It creates the index database.Put the established index into alocal file. Then carry out the search though retrieving the index database.Because ofthe powerful search function, it costs little time in massive data searching and canmeet the requirements of real-time entity search in Internet of Things. Althoughreal-world information access got the information of entities real-timely, but theinformation in Internet of Things are of highly dynamic and the data areinstantaneity.To implement the real-time search better, the prediction module behindthe search module was extended. Cycle prediction models, includingAPM-Aggregated Prediction Model, SPM-Single-period Prediction Model,MPM-Multi-period Prediction Model were used. By judging the event cycle model ofInternet of Things, reasonable prediction model was used. Predict the probabilityvalue of the state of entities in Internet of Things can make the real-world entitysearch more instantaneity.In order to verify the instantaneity and accuracy of searching and predicting for real-world entities, the sensor data sets of the U.S. Mitsubishi Electric ResearchLaboratories and Minnesota traffic data sets were used for experiments. Experimentsshow that search for accurate results from massive entity data in a short time isachievable.It can do a good instruction for the users’ decision.