Research On Key Techniques Of Search In The Internet Of Things

As in the traditional Web, search is also a key service in the Internet of things.It is very difficult for search in the Internet of things to realize the real time search of the designated state of sensors because of the dynamic nature of the sought state.How to return the result set of sensors that match the sought state with high efficiency and low time delay at the presence of huge sensors and dynamic state is the primary problem to be solved in this paper.Combining the algorithm of mining obscure period in data mining and regularity of people’s behavior in daily life, this paper presents a prediction model based algorithm of sensor ordering,which efficiently implements the sensor state based sensor search. The basic idea is to compute a ranked list of sensors with the use of history data sensors outputted such that the higher the rank of a sensor in this list, the likelier this sensor matches the query. This way, a search engine can process sensors in the order of their rank, spending effort on sensors first that are most likely to match the query. Such effort includes in particular reading the current output value of the sensor over the Internet to check if it actually matches the query. Without sensor ranking, It will spend huge effort on sensors that are unlikely to match the query anyway, and reduces the latency of producing the result pages. In this paper, the classroom online booking system dataset (ORS) and office building mobile monitoring sensor dataset (OMD) validate sensor ordering algorithm of multiple prediction models for the scene and advantages.In addition, it is a difficult problem to be solved in this paper to accurately represent the prediction model and integrate it with the existing search engine. Based on the currently development of Web and semantic sensor web ontology, the RDF resource description framework and the SPARQL query language, and the open source framework Jena and ARQ, a variety of prediction model of the semantic description are presented. To improve the model in the basic semantic forecasting model in the module additive factor method, gradually adding time, place and relationship factors, the paper finally completed the a simple and easy-to-use sensor query system, enabling the prediction model with the search engine to complete the integration.Users can get the result set of sensors by designating the time,place and state of sensors.