Seite - 172 - in Intelligent Environments 2019 - Workshop Proceedings of the 15th International Conference on Intelligent Environments

takes place by exchanging information.With these criteria in mind, we are automati- cally in thefield of distributed computing as information is exchangednot only locally, i.e. within the boundaries of a component, but also cross-component. Since Internet of Thingssetupsarenaturallydistributedsystems,as the“Things”arecomponents interact- ingwith eachother by exchanging informationover the “Internet”, the citeddefinitions are sufficient even for theareaof IoT. 3. ModelsandApproaches forHierarchical InteroperabilityMeasurement In section 2, interoperability has been identified as ameasurable property indicating to whatdegree interoperability isgivenbetweentwoormoreparticipants.Theworddegree implies thatweneed some type of scale or a classification approach such that different interoperability scenarios can be comparedwith each other and ranked. Persisteras and Tarabanis have analyzed twelve proposed classification approaches grouping together different aspects of interoperability [6]: The finding of their analysis is, that all twelve classification approaches use an evolutionary perspective,whichmeans that the degree of interoperabilitycanbemeasuredbyansweringthequestionwhichaspects (alsocalled features [6]) of interoperability are fulfilledandwhichnot.The fulfilmentof an specific aspect in turn, canbemeasuredby the fulfilment of criteriawhich are defined in accor- dancewith the aspect. Some aspects aremore advanced than other (and therefore their criteria areharder to fulfil) leading toanhigher, i.e. amoreadvanced,degreeof interop- erability. For instance, the fact that twoparticipants are able to exchange symbols over awire is a lower fulfilled aspect in comparison to the aspect, that these twoparticipants agree on the same understanding of themeaning of exchanged data structures. These aspectsand their criteriaareconsolidatedwithinso-called levelsof interoperability such that amodel for classification typically consists ofmultiple interoperability levels fol- lowingastrict linearity:Forreachinganupper levelof interoperability(i.e.ahighdegree of interoperability) all the aspects of the underlying levelsmust be fulfilled.Although, thenumberof interoperability levelsvary fromapproach toapproach, theaspects identi- fiedby[6]within the twelveanalyzedclassificationapproachesareoftensimilar toeach other: On the lower levels, the basic aspects of communication are addressed, such as the ability to exchange single symbols up to unstructured and structured data (e.g. in [11] and [12]),whereas thehigher levels covers aspects like the semantic representation of singlewords, data structures up to the ability to integrate the provided services of a component into anworkflow for achieving a certain goal (e.g. in [11], [13] and [14]). AlthoughPersisterasandTarabanishaveanalyzedthese twelveclassificationapproaches in 2006 and in the context of information systems in general, classification approaches and typologies for IoT interoperability that havebeendeveloped later use the sameevo- lutionary perspectivewith different interoperability levels asmentioned above. Recent models andapproaches for IoTcanbe found in [3] and [7]. Dueto thescaleof fulfilmentornon-fulfilmentofaspectsandtheircriteria, themea- surementwithin thesemodels is hierarchical andnot basedonametric basis.However, a long-termgoal shouldbe tomeasure, evaluate andcompareparticular interoperability scenariosnotonlyhierarchically,butalsobasedonametricscale inorder toachievefiner granularity. As alreadymentioned, our contribution is to offer such ametricmeasure- mentmethod.However, thismethod is limited to only one specific aspect, namely the S. Kotstein and C. Decker /AnApproach for Measuring IoT Interoperability172