Research On Semantic-based Service Architecture And Key Algorithms For The Internet Of Things

The Internet of Things（IoT） had formed a service network system，which was integratedof all kinds of application based on sensing technology. Its core value was―Smart Service‖.Most of the IoT services were faced with some dynamic adaptability problems, such asmorphology changes, denotation extension, environmental changes, and business adjustment.They were also faced with different levels of sharing and interoperability. In addition, thenature of magnanimity and heterogeneous of IoT services made their integration andextension be faced with new challenges. Thus, there was an urgent need to build a serviceplatform based on the semantics to make large-scale integration of the ubiquitous andheterogeneous services. Multiple levels of semantic support would be needed to meet theneeds of the common understanding and interoperability of the services. Finally it wouldachieve the efficient discovery and personalized delivered in this platform.The current related researches mainly concentrated in the close-loop aspects in a certainlocal area, such as sensing recognition, communication protocols, application realization andso on. Pay little attention to open-loop aspects in cross-domain, such as resources addressing,service mode, service providing mechanism and so on. In particular, the related researches onIoT service platform were still in the germination stage of theory exploratory. For example,there were little standards for IoT service platform architecture until now, and the constructionof ontology and the semantic description in IoT service area which was still in its infancyneeded to be improved, and the key strategies of the platform were not mature enough. Webservices technology provided a valuable reference to solve the sharing and integration of IoTservices, but it was difficult to adapt to the characteristics of IoT services. For example, asingle registry mode couldn’t satisfy the demand of IoT service platform to be expanded intoubiquitous environment. The performance of traditional centralized service discoveryalgorithm in computing the semantic matching was poor in large-scale service. Most of theP2P service discovery algorithm was difficult to adapt to network changes drastically and theinstability of communication link. The process of service selection did not consider for usercontext adaptation, so it was difficult to meet the demand of the user’s personalizedexperience, etc. In order to solve the above problems and mollify the contradiction betweenthe urgent demand of IoT services support platform for the rapid growth of IoT service andthe lack of the related research, the following work had been done in this paper.1. IoT service platform architecture based on semantic technology was constructed.Based on the comprehensive analysis of the characteristics of IoT service and service platform, following the existing standard of some researches, IoT service platform based onsemantic technology and its architecture was proposed. This platform extended the servicesource from the generic services in UDDI registry to the service of self-organization inubiquitous environment, to support the needs of service sharing and interactive for each level.2. IoT Service ontology model was constructed.In order to meet the requirements of the unified sharing and flexible expansion of therelated ontologies for IoT services, we proposed a three-layer IoT ontology model, whichincluded the generic ontology layer, the core ontology layer and the application ontology layerfrom the top to the the down. In this structure, the lower ontology put forward the expressingneed to upper ontology. And the upper ontology provided the lower ontology with the designprinciples. Also we gave the modeling of each specific ontology in core ontology layer toguide the modeling of service application ontology in the field of various industries.3. Semantic-based IoT service description language OWL-S^iotwas presented.By contrasting the existing typical Web service semantic description language, OWL-Swas selected as a foundation for the evolution of language for its better adaptability andwidely applied. In order to meet the dynamic flexibility requirements of IoT servicedescription, by some professional processing for OWL-S, a suitable description language forIoT service called OWL-S^iotwas formed. The ontology and the OWL-S^iotwould provide thenecessary semantic support for IoT service platform.4. Respectively gave the service management strategy both in registry mode andself-organization by sub-cluster.IoT service contained both the traditional common services, and the new ubiquitousservice. So it needed to be managed respectively. For the former, we gave the service registryarchitecture based on semantics. For the latter, in order to save energy consumption of nodesand do not significantly affect the connectivity of the network, we put forward aself-organization management strategy using sleep scheduling based on―game of life‖.Finally we gave the experiment to analyze the impact of the service interaction delay underthis strategy.5. A functional computing and context-aware computing strategy to explain thefunctional requirements and the context of the user was introduced.Functional computing mainly used the technology of the core word segmentation, thedomain dictionary matching and the interactive confirmation method to explain use’sfunctional requirements which was inputted in natural form. In particular, for the compositedemand, a decomposition method is presented. Context-aware computing mainly used acontext fusion model based on genetic algorithm and decision tree to identify the originalcontext, in order to obtain the high-level context which could be explained. Through acomprehensive method of calculating vocabulary semantic similarity based on semanticdictionary, achieved the semantic annotation of functional requirements and context based onthe related ontology to generate the description document of user’s requirements automatically.6. A more efficient algorithm of service discovery was proposed.For the traditional generic services which were organized in registry, in order to dealwith large-scale service discovery, a centralized service discovery algorithm based onmulti-stage services semantic matching was proposed. According to the characteristics of eachstage, we respectively presented the similarity computing algorithm for the single concept, themultiple concepts and QoE. For the decentralized autonomous services in ubiquitous,proposed a P2P service discovery algorithm based on Markov prediction model of meetingtime span. The markov chain was used to achieve the predicte prediction of the time span ofthe meeting of the nodes, and used the model to improve the Spay and Wait routing protocolto improve the delivery success rate of service packet. The experiments showed that both ofthese algorithms improved the recall rate and precision in small delay.7. A service selection algorithm based on context and service rank was proposed.In order to avoid blindness and randomness of service selection, we proposed afine-grained and personalization service selection algorithm. Not only considered thepersonalized demand for user’s context, also considered the fine-grained demand for servicelevel. Put forward a calculation method of the coordination of semantic and numerical, tosolve the matching problem between the heterogeneous context parameters. Also theestimation method of service rank based on fuzzy logic was given. Finally, the experimentalresults verified the validity of the algorithm.To sum up, this work could enrich the research about IoT services architecture andservice mechanism, and meet the needs of real-world applications of IoT services integration.It had certain theoretical significance and application value.