D5.2.1 Decentralized and Autonomous Things - iot
Contents
1. Acronym Meaning API Application Programming Interface CB Case Base CBR Case Based Reasoning D Deliverable FM Friends Management GUI Graphical User Interface HTTP Hypertext Transfer Protocol ID Identifier IDE Integrated Development Environment loT Internet of Things JDK Java Development Kit KB Knowledge Base K NN K Nearest Neighbor OS Operating System OWL Web Ontology Language PPM Profiling and Policy Management RDF Resource Description Framework REST Representational State Transfer SA Social Analysis SM Social Monitoring SNA Social Network Analysis SPARQL SPARQL Protocol And RDF Query Language TTL Time To Live Date 22 07 2014 Grant Agreement number 609043 Page 5 of 24 D5 2 1 Decentralized and Autonomous Things Management cosmos URI Uniform Resource Identifier VE Virtual Entity VM Virtual Machine WP Work package XP Experience Date 22 07 2014 Grant Agreement number 609043 Page 6 of 24 D5 2 1 Decentralized and Autonomous Things Management 1 Introduction This document is the complement to the delivered software regarding the month 10 WP5 prototype deliverable More specifically the structure of the document is as follows e Sections 2 4 Discusses the services that have been developed till now under the several components of WP5 The schema of the analysis followed in each one of these sections is o Description Provid2. D5 2 1 Decentralized and Autonomous Things Management cosmos COSMOS Cultivate resilient smart Objects for Sustainable city applicatiOnS Grant Agreement N9 609043 Version Due Date Delivery Date Nature Dissemination Level Lead partner Authors Internal reviewers D5 2 1 Decentralized and Autonomous Things Management Software prototype Initial WP5 Decentralized and Autonomous Things Management 1 0 30 June 2014 22 July 2014 Prototype Public 4 ICCS Orfefs Voutyras ICCS Panagiotis Bourelos ICCS Adnan Akbar UniS Date 22 07 2014 Grant Agreement number 609043 Page 1 of 24 D5 2 1 Decentralized and Autonomous Things Management www iot cosmos eu The research leading to these results has received funding from the European Community s Seventh Framework Programme under grant SEIT TRACER agreement n 609043 PROGRAMME Version Control Version Date Author Author s Organization Changes 0 1 11 7 2014 Orfefs Voutyras ICCS Sections 5 6 7 8 Panagiotis Bourelos introduced 0 2 14 7 2014 Orfefs Voutyras ICCS Providing more content Panagiotis Bourelos for sections 2 3 4 0 3 18 7 2014 Orfefs Voutyras ICCS Figures added Panagiotis Bourelos 0 4 21 7 2014 Orfefs Voutyras ICCS First version for Panagiotis Bourelos internal review 0 5 22 7 2014 Orfefs Voutyras ICCS Final version after UniS Panagiotis Bourelos review Anne
3. Flat2 VE Flat3 VE alt J Case Presb Locally 1 Notify of Alrrival tagret Time target Temp Confirmatign message Case Present in Flat2 1 1 Boiler Temp URI Messa F Case Present only in Flat3 3 Notify of 2 Notify of Alrival tagret Time target Temp rival tagret Time target Tem Search 2 1 1 Boiler Temp URI Mess 3 1 Search Sifi 3 1 1 Boiler Temp URI Mess Confirmati nlmessage S 2 1 1 1 Boiler Temp URI Mess ilar Case Temp Before Time Temp After 3 1 1 1 Boiler Temp URI Message E that answered Request XP gg Before Time Temp After imilar Case Temp Before Ti 3 1 1 1 1 Request XP me Temp After After H Similar Case Figure 6 Sequence diagram of the CBR and cases discovery mechanism Date 22 07 2014 Grant Agreement number 609043 Page 21 of 24 D5 2 1 Decentralized and Autonomous Things Management sd Simulation of loT service Discovery J ArrayList lt gt Input Request Service Description Domain Target VE Service Name URL Input Parameters Output Parameters 1 1 ArrayList lt gt Input Retrieve Services D Figure 7 Sequence diagram of the loT service discovery mechanism sd Friend List Renewal J 1 New List of Friends Requdst Friend List Renewal Old List of Friends Threshold Weights
4. e Input Problem parameter names Problem parameter values Solution parameter names Address IP and port ttl e Output Solution parameter values Similarity percentage URI Message Who gave the solution Date 22 07 2014 Grant Agreement number 609043 Page 10 of 24 D5 2 1 Decentralized and Autonomous Things Management cosmos 3 2 2 loT services discovery The DiscoverloTService mechanism is initiated by the discoveryMechanism method that contacts a remote VE through the Http POST method in order to extract information about the semantic description of services pertaining to a specific domain When the remote VE is contacted it uses the Planner s searchloTService method as described below to provide the answer No recursiveness has been added into the mechanism yet Below is a brief outline of the input output parameters of the mechanism e Input domainName IP address amp port e Output Service name amp URI Names of input parameters of the service Names of output parameters of the service The searchloTService implemented in the Planner component uses the input provided by the discoveryMechanism method above more specifically the domain name and retrieves from the service local store the semantic description of any relevant loT services The answer is a structured ArrayList variable containing all relevant data as stated above 3 2 3 Friends discovery The requestMoreFriends method initiates the Friend Disc
5. The Planner will become part of the VEs during their registration time and will run locally In order to measure similarity between cases we need to follow two steps First of all the similarity of the attributes of the new case with the various cases in the CB has to be measured This is done in the source code by using the method compareProbParams in the Planner class This method takes as input the names of the parameters of a new incomplete case a case that consists of a problem and no solution and the percentage of acceptable similarity The similarity is calculated by using the Jaccard coefficient If the compareProbParams returns as an answer true meaning that there are cases in the CB that have sufficiently similar attributes to those of the new case the searchSimilarCase in the Planner class is calculating a new similarity index in order to find the most appropriate case This method takes as input a list of the names of the attributes describing the new case a list of the values of the attributes a list of the weights of these attributes if any and a new similarity threshold regarding the values of the attributes and returns the most appropriate solution values of the attributes of the solution a URI or a message In this case the similarity is calculated by using the Bray Curtis distance It should be noted at this point that the Planner has two modes Whenever a VE creates a new case on its own the searchSimilarCase method
6. The user should be able to add friends to his her VE manually too 3 During the creation of an application COSMOS offers the developer the opportunity to define how cases relevant to his application are going to be stored created and maintained 4 Auser downloads an application for a VE This application defines how the several cases both complete and incomplete are going to be created For example the flat VE continuously records the actions of a human user regarding the heating of the flat In our case it may record that at a room with a temperature of 6 C the user turned the heating on for a total of 10 minutes 600 seconds and stopped at 20 C using hot water of 70 C 5 That way the VE builds a case of 6 600 20 70 and in a similar fashion its case base COSMOS manages the CBR cycle 7 Each time a new incomplete case is created the VE searches its Case Base for a solution For example the user may inform the VE that he she will return after a specific time interval and request a certain target temperature in that certain time limit 8 f no solution is detected the VE initiates the Experience Sharing service and asks its friends for help 9 Anumber of cases is returned by the friends 10 Based on the dependability index of the friends and the similarity of the cases the weights of these criteria can be defined the cases are sorted and the best case is chosen 11 The case is evaluated based on its results and
7. e Executables o target DemoProjectMaven 1 0 SNAPSHOT jar target ExperienceGiverVEMaven 1 0 SNAPSHOT jar target JennaPelletQuerryMaven 1 0 SNAPSHOT jar target SocialAnalysisComponent 1 0 SNAPSHOT jar target FriendOfFriend 1 0 SNAPSHOT jar OOoOo0O O0 7 2 Download and Installation instructions The development took place in an x64 Windows 7 OS The prerequisite for the execution of the applications is JDK 1 8 This is bundled along with the NetBeans 8 0 IDE that was used for code development The entire bundle can be accessed at http www oracle com technetwork java javase downloads index html After downloading and installing the previously mentioned components the next step is to transfer all Project Folders to the newly created NetBeansProjects Folder inside the Documents section After starting NetBeans and opening the Projects that use Maven for greater ease in transportability between workstations the user may be asked to execute resolve problems so that all dependencies are imported By right clicking on the project names and selecting resolve problems Maven will connect to the central repository and download to a new local repository all dependencies for current and future use Also it is important to add that in order for the projects to run correctly the accompanying localstore folder must be placed as is in the main hard drive of the workstation Date 22 07 2014 Grant Agreement number 609043 Page 16 of 24 D5 2
8. is used However if a VE accepts a request for the search of a new case from another VE the searchSimilarCaseRequest method is used In the last case the reasoner searches only for shareable cases defined by flags These flags can be set either by the application developer or the owner of the VE To conclude in our case the simple flat memory k Nearest Neighbor retrieval K NN is used In this approach the assessment of similarity is based on a weighted sum of features Date 22 07 2014 Grant Agreement number 609043 Page 9 of 24 D5 2 1 Decentralized and Autonomous Things Management cosmos 3 Decentralized Discovery Mechanisms 3 1 Description The concept of adapting past solutions is one of the main requirements of CBR For this reason we introduce the idea of allowing VEs to share their cases thus producing an environment where the knowledge is distributed and knowledge flow is supported In that sense the cases produced via CBR become one form of Experience of the VEs which can not only aid in the planning of future solutions but can also be shared between different VEs Because of the great distribution of the knowledge among the VEs the development of decentralized discovery mechanisms becomes essential For this reason we use mechanisms such as ttl time to live in order to support a controlled flow of requests among the VEs As a result the first steps towards decentralized discovery of Cases loT services a
9. 1 Decentralized and Autonomous Things Management 7 3 User Manual Though the applications can be run by their accompanying executable jar files inside the respective target directories of the Project files it is best to run them through NetBeans This way the user can observe all underlying messages in the output window of the IDE which would not appear during execution just by jar The first component to be demonstrated is the use of the Planner responding to a notification pertaining to the scenario of Efficient Heating as described in section 5 The first phase should be the running of the ExperienceGiverVE and JennaPelletQuerry Projects as they simulate two auxiliary VEs that participate in the experience sharing process that may be deemed necessary by the values given during the demonstration Finally the user must run the DemoProject in order to start a GUI that provides the means to begin the process The Demoproject simulates the GUI waiting for user input as well as the primary Flat VE receiving the notification fm Notification Area Variables Return time in seconds Desired room temperature localhost Flat1 vj IP address of my house Portthat my house listens on Clear Send Notification 3030 Load LocalStore Create Case Notification unkwown time Figure 1 Main interface of the GUI By adding the values 3535 in the first box and 26 in the second box we can begin the notificati
10. 4 5 2 MUOPERLSOODAEIO uo novit n anum RMAGASURU rep East RI teas DEF N a REUNIR RUE ose aaa ie 14 6 Used ontology fes ttes tette Lati aie ii Een eaa aa T RAE 15 T Deliyery and WS BS sossa E T 16 TAs PARAS DIODRSBOHL edd esiste arrea naain nenii iiai 16 7 23 Download and Installation instructions eene 16 MEE CAU DcN m 17 T4 Licensing rire TOSS REO iss kiasi eiia 20 PEE esiti entium c E R E 21 Date 22 07 2014 Grant Agreement number 609043 Page 3 of 24 D5 2 1 Decentralized and Autonomous Things Management cosmos Table of Figures Figure 1 Main interface of the GUI 17 Figure 2 Message returned after a successful case retrieval from the local case base 18 Figure 3 Case returned from Flat2 18 Figure 4 Case returned from Flat3 with Flat2 acting as a broker 18 Figure 5 The result of a friend recommendation service provided by a friend of the initial VE 19 Figure 6 Sequence diagram of the CBR and cases discovery mechanism 21 Figure 7 Sequence diagram of the loT service discovery mechanism 22 Figure 8 Sequence diagram of the Friend List Renewal mechanism 22 Figure 9 The components used from the CBR and cases discovery mechanisms 23 Figure 10 The components used for the social links establishment 23 Date 22 07 2014 Grant Agreement number 609043 Page 4 of 24 D5 2 1 Decentralized and Autonomous Things Management cosmos Table of Acronyms
11. COSMOS Social Analysis Component 1 1 Calculate Depenability Indexes Old List of Friends Weights 2 Friend List File Qr gt l lt 1 2 Build New List Dependability Indexes Threshold cal Friend List New Friend List Date 22 07 2014 Figure 8 Sequence diagram of the Friend List Renewal mechanism Grant Agreement number 609043 Page 22 of 24 pk D5 2 1 Decentralized and Autonomous Things Management Finally the conceptual view of the main cycles described above CBR discovery mechanisms and social links establishment are depicted in the following figure case foung VE3 VE2 A case not found social info Figure 10 The components used for the social links establishment Profiling and Policy Management PPM component It assigns a unique ID to the VE and enables the entry of all the information needed for the description of the physical entity through the domain ontology of the corresponding VE Moreover it enables the owner to determine the social openness of the VE the loT services that can be used by other VEs the kind of experience that can be shared the sets of VEs which can access such information etc Friends Management FM component It is responsible for creating and maintaining the list of friends that a VE has In other words it allows VEs to initiate update and terminate their friendship with other VEs on the basis of the owner s contro
12. cing various situations This is a step towards the autonomicity of the VEs as depicted by the goals of Task 5 2 Autonomous and predictive reasoning of things The reasoning technique that was chosen for the development of this component is the Case based reasoning CBR CBR is the process of solving problems based on past experience In more detail it tries to solve a case a formatted instance of a problem by looking for similar cases from the past and reusing the solutions of these cases to solve the current one Cases encompass knowledge accumulated from specific specialized situations The advantages and disadvantages of case based reasoning and the reasons we chose this reasoning technique instead of others were analysed in COSMOS WPS D5 1 1 pp 17 19 The CBR reflects human reasoning and is the easiest approach for the developers and the one that can be used in a huge variety of domains Application developers will have the opportunity to create new applications based on CBR by defining how new cases should be created what is going to be the pattern of the problems and the solutions COSMOS from its side will provide the VEs which will download these applications with the necessary reasoner and all the mechanisms needed for the CBR cycle Section 4 1 5 in COSMOS WP5 D5 1 1 2 2 Functionalities 2 2 1 Case Base The application developers will be able to define how new cases should be created by describing them through an on
13. d in the calculation of the new Dependability Indexes by the Social Analysis component Once the component receives the input from the initiateFriendListRenewal method it will use the RESTful input to calculate Dependability Indexes for existing friends through the method described below After this step the Social Analysis component will make use of a threshold of Dependability currently hard coded in order to purge the list of no longer reliable friends If indeed VEs have been purged then the Social Analysis will proceed to calculate new Dependability Indexes for other similar VEs inside the Social Ontology After sorting the returned VEs by their Indexes it will append the data of the most dependable one until the original friend list size is reached Finally the data returned to the original VE will then be used by the buildFriendFile method in order to recreate the friend list with the new friends and their new Dependability Indexes Date 22 07 2014 Grant Agreement number 609043 Page 12 of 24 D5 2 1 Decentralized and Autonomous Things Management cosmos 4 2 2 Dependability Index calculation Having described the process through which a VE can contact the COSMOS Social Analysis component and have its friend list renewed it is at this point wise to actually describe the way new Dependability Indexes are calculated At every VE the stored Dependability of its friends is subjective meaning that although it is derived b
14. eir short descriptions and licences as follows e APACHE JENA 2 10 0 Apache Jena is a API and toolkit for working with semantic web technologies such as RDF and SPARQL using Java This artefact represents the source and binary distribution packages generated for releases Licenses Apache 2 0 License http www apache org licenses LICENSE 2 0 e Jetty Maven Plugin 9 1 5 v20140505 Administrative parent pom for Jetty modules Licenses Apache 2 0 License http www apache org licenses LICENSE 2 0 e Pellet Jena 2 3 2 The Clark and Parsia Pellet OWL Reasoner Licenses GNU Affero General Public License 3 0 http www gnu org licenses agpl 3 0 html e Maven Dependency Plugin 2 8 Provides utility goals to work with dependencies like copying unpacking analyzing resolving and many more Licenses The Apache Software License Version 2 0 http www apache org licenses LICENSE 2 0 txt e Maven JAR Plugin 2 4 Builds a Java Archive JAR file from the compiled project classes and resources Licenses The Apache Software License Version 2 0 http www apache org licenses LICENSE 2 0 txt Date 22 07 2014 Grant Agreement number 609043 Page 20 of 24 D5 2 1 Decentralized and Autonomous Things Management 8 Architecture Below there are some sequence diagrams that represent some of the mechanisms described in the previous sections sdFirst Simulation of Experience Discovery and Planner Use Home Owner VE of Mobile Flat VE
15. es the main description of the component the rationale behind deciding its development and general advantages that it can offer to end users o Functionalities Describes briefly the main services that are currently provided by the component while using references to the source code e Section 5 Describes the use case that has been chosen in order to demonstrate the use of the main components and services of WP5 e Section 6 Contains a brief description of the ontology files that have been produced in order to provide input to the several services described in Sections 2 4 e Section 7 It provides delivery and usage information regarding the prototype source code and is organized as follows o Package information o Download and installation instructions o User manual The most important part of this document It describes how the results of the several services implemented by the source code can be showcased even in a small degree o Licensing information e Section 8 Summarizes the various relations between the different components and services of WP5 that have been presented on this document Date 22 07 2014 Grant Agreement number 609043 Page 7 of 24 D5 2 1 Decentralized and Autonomous Things Management cosmos 2 Planner Case Based Reasoning 2 1 Description The main functionality of the Planner is providing the VEs the ability to run applications by using a reasoning technique for solving problems and fa
16. ex are stored in the datatype properties of hasReputationIndex hasReliabilitylndex and hasTrustIndex of each VE The only pre requisite is that by definition the reputation index must be equal or less than the trust index and the reliability index is a float number between 0 1 Note that all data in the ontology is stored as strings The file names used are friendlist owl and CosmosSocial owl inside the localstore folder Finally the Project FriendOfFriend implements the Friend Discovery mechanism described previously in a basic way at the moment it is self called and executed so that a VE can request other friend VEs for friend recommendations based on their own locally stored Dependability Indexes In this case the target VE returns a name or possibly other identifier of its best friend so that the original VE will be able to receive its data and add it to the friend list after communicating with the SA component The code implements the return of the answer from the target VE with further functionality to be added n E Flat15 if you want to end the program click the OK button Figure 5 The result of a friend recommendation service provided by a friend of the initial VE Date 22 07 2014 Grant Agreement number 609043 Page 19 of 24 D5 2 1 Decentralized and Autonomous Things Management cosmos 7 4 Licensing information The libraries used in the development of these prototypes are listed along with th
17. h different file names All files used for storage and retrieval of data are owl and their general structure and purpose is as follows e List of Friends The list of friends is a collection of individuals along with the local VE and their data type and object type properties i e hasFriend or hasDependabilityIndex The files are named friendlist owl friendlist1 owl and friendlist2 owl with the first belonging to Flat1 the second to Flat3 and the user VE shared and the third one to Flat2 e List of Cases Case Base CB These owl files are used for the storage and retrieval of data pertaining to the information that is described by Cases Cases are represented as two separate individuals problem and solution connected through the use of the solves and isSolvedBy inverse functions They also possess data type properties containing their relevant data The files are named casebase owl casebasel owl and casebase2 owl with Flat1 owning the first Flat3 the second and Flat2 the third e Service stores These files that may or may not be merged with the rest dependent on the course of the code development contain the semantic description of loT Services and are used for the DiscoverloTService mechanism of our prototype They are VE owl VE1 owl and VE2 owl following the same pattern of ownership e Cosmos Social Ontology This owl file contains all relevant data for VEs that can be used to infer social connectivity between them and e
18. hanism Case Base Discovery Another functionality that is implemented is a first draft of the Social Analysis SA component expected to run on COSMOS and more specifically its function as a means to provide a renewal mechanism for the friend list of individual VEs based on the information of the COSMOS Social Ontology This code is implemented in the SocialAnalysisComponent Project which simulates a VE sending the names of its friends to the Platform in order to recalculate their Dependability Indexes The SA uses the VE provided weights to calculate the new Index based on the reputation trust and reliability that each VE has inside the Social Ontology After checking the new Dependability Indexes with the threshold set by the requesting VE and if this has led to the purge of friends from the original list it gives an analogous number of new friends that are above the threshold probabilistically which means not ordered by the new index RebuildFriendList mechanism After the entire process has been completed it returns the new friends to the VE along with relevant data so that the VE can rebuild its friend list ontology buildFriendFile method Any changes in the friend list can be observed by using Prot g to view the friend list owl file and can be triggered by modifying values in the Social Ontology owl file that the component accesses to recalculate the Dependability Index Namely the values that are used to calculate the Dependability Ind
19. l settings It provides the owner with the option of setting new friends to his her VEs offers friend recommendation request services and monitors the friends list of a VE regularly or on demand in order to find any Friends whose Dependability is no more the desired one and thus should be removed For this purpose it communicates with the SA component Date 22 07 2014 Grant Agreement number 609043 Page 23 of 24 D5 2 1 Decentralized and Autonomous Things Management Social Monitoring SM component It contains all the main tools and techniques that are used for the monitoring of the social properties of the VEs like Trust and Reputation Its main objective is to collect aggregate and distribute monitoring data events across the decision making components of the collaborating groups The events are generated by interactions in response to directly or indirectly user actions e g registering a new VE or VEs actions XP sharing Social Monitoring feeds the VE Registry Social Analysis SA component Based on the results of the Social Monitoring component and taking advantage of Social Network Analysis SNA the SA component is used for the extraction of complex social characteristics of the VEs e g centrality as well as models and patterns regarding the behaviour of the VEs and the relations between them Date 22 07 2014 Grant Agreement number 609043 Page 24 of 24
20. nd Friends are made 3 2 Functionalities 3 2 1 Cases discovery The method experienceSharePOST is the starting point of the Case discovery mechanism It uses the Http POST method to pass data into a remote VE targeting the URI of the Experience Sharing mechanism The method structures the request into a POST body by creating a String variable containing the required variables and attributes Upon receiving the request the remote VE extracts the data and checks the value of the ttl number If by reducing it the ttl number is above zero then it calls on the suitable Planner Component methods in order to check if a similar problem structure exists inside the local CB compareProbParams and after a positive reply initiates the searchSimilarCase method of the Planner If a similar case is not retrieved then the remote VE becomes a broker by initiating recursively a new call of the Experience Share mechanism using the new ttl calculated by the method described in 3 2 4 of this document Thus we ensure that the entire process will operate until either a suitable solution has been discovered or until it reaches a dead end ttl time out no case present in our friend VE cluster It is worth noting that in the case of outward generated case retrieval the VE will only retrieve cases marked as shareable in the local CB by the isShareable data type property Below is a brief outline of the input and output of the Experience Sharing mechanism
21. ned case is 100 096 Finally the URI of the recommended service is blah setthermostatitemperature Figure 4 Case returned from Flat3 with Flat2 acting as a broker It is worth mentioning that during this example the output tabs inside NetBeans will at some point display the output of the DiscoverloTService mechanism that Flat1 implements in a basic Date 22 07 2014 Grant Agreement number 609043 Page 18 of 24 D5 2 1 Decentralized and Autonomous Things Management cosmos fashion The service contacts a VE itself at this example and requests the values of the semantic description of any service associated with the domain provided as input Until this point the Social Monitoring s function was to streamline the communication between VEs by choosing the appropriate friends from the local friend list As described another main target for the component is to provide a mechanism to dynamically change the depth of the recursiveness in Discovery Services based on the number of friends popularity of intermediary broker VEs Such a mechanism will call on the component s getMaxTTL function that retrieves the previously calculated value and during the reception of incoming REST communications will modify the incoming ttl if it is deemed necessary Also the calculation of the max value will take place ideally after each friend list renewal In our example such checks for the maxttl variable are made during the Experience Sharing mec
22. on process as soon as we hit the Send Notification button At that moment the application sends an http POST request through the sendNotification method inside the Servlets java file to the recieveNotification Service implemented inside the same file After the POST I received the VE calls an instance of the Planner and begins searching for a similar case as the one provided by the user The case structure is TempBefore time TempAfter and boilerTemp URI of Service message Current temp is always set at 5 degrees The searchSimilarCase method will run without first checking the compareProbParams function as it is expected the VE will contain such a structured case by design With the provided values the VE searching the local case base will indeed find a similar case and thusly return Date 22 07 2014 Grant Agreement number 609043 Page 17 of 24 D5 2 1 Decentralized and Autonomous Things Management The boiler has been set at 60 Celcius The Service to be contacted is blah setthermostat temperature The Message to be displayed is None Figure 2 Message returned after a successful case retrieval from the local case base Afterwards we enter the values 1800 and 26 Such a case is not present inside the first VE so after failing to retrieve it the VE will call the experience share service and contact through http POST its friend Flat2 implemented by the ExperienceGiverVE project The second VE after receiving the re
23. overy mechanism in a VE with the input being only the address of a remote VE which will provide us with a suitable friend candidate The philosophy of the previous discovery mechanisms is used meaning RESTful communication through Http POST When the remote VE receives the request and begins to process it in its doPost method part of the same FriendRecommendation class it calculates its most subjectively dependable friend based on its own subjective Dependability Index and returns the information necessary as a recommendation to the original VE Further code development will include the ability of the original VE to ask the Social Analysis Component of the COSMOS platform for relevant data of the recommended new friends so that it can establish new social links 3 2 4 Calculation of max ttl For each one of the discovery mechanisms described previously one of the main goals is refining of their recursive abilities A big step in this direction is the use of the ttl variable not only as a static although customizable input but as an actual dynamic representation of the in between stages of information brokering based on the theory of the six degrees of separation and the actual social capabilities of affected VEs This means that if a ttl begins high enough and the recursive discovery uses nodes broker VEs with many outward connections this can have an adverse effect on network overhead and discovery overlap Therefore by using the calc
24. quest will also call on its Planner in order to check the structure of the received experience request If the compareProbParams method returns true then and only then it will proceed to access its case base in order to locate a similar problem solution pair After locating such a case the Experience Sharing Service returns to the first VE which displays the answer as follows Message Ex Oops no case like that better check with my friends just did and will setthe boiler temp at 54 degrees Celcius The friend that helped me has the Hardcoded Name Flat2 The similarity of the returned case is 100 096 Finally the URI of the recommended service is blah setthermostattemperature Lox J Figure 3 Case returned from Flat2 Finally in order to check the recursiveness of the Experience Sharing Service we will have to enter the values 3535 and 28 in the GUI fields Such a combination of values is present in neither Flat1 nor Flat2 so after failing to find a similar case Flat2 will call Experience Share recursively on Flat3 implemented by the JennaPelletQuerry project While searching the local CB Flat3 will in turn locate a similar case and return it to Flat1 VE with Flat2 acting as a broker between the two of them Message Ea Oops no case like that better check with my friends just did and I will set the boiler temp at 69 degrees Celcius The friend that helped me has the Hardcoded Name Flat3 The similarity of the retur
25. the Trust of the friend that shared its case is recalculated 12 Atsome point the VE may want to refresh the list of its followees friends 13 The VE sends a friend recommendation request to the Social Analysis component 14 The parameters passed to the request include weights for calculating the dependability trust and reliability a minimum acceptable limit of it and the list of followees the VE already has 15 A new friend list is returned that includes the dependability indexes Friends that have been set by users should not be thrown away but isolated 9o 5 2 Madrid Scenario In this use case all buses are exactly the same have the same owner and send all of their data centrally to EMT As a result the development of a nice scenario for showcasing important mechanisms of WP5 like the flow and management of the VEs knowledge was proved to be really difficult and for this reason we focused during year 1 on the London scenario Date 22 07 2014 Grant Agreement number 609043 Page 14 of 24 D5 2 1 Decentralized and Autonomous Things Management cosmos 6 Used ontology files For the prototype code s needs we make use of the ontology files contained inside the localstore folder that accompanies the project files Because of the fact that we do not yet have access on different VMs to test the code that has been developed until this point we use the same file storage structure for each simulated VE but wit
26. tology Generally a case will consist of i a problem which is going to be a series of events that have to be identified to trigger the solution This description of events has to be linked with the corresponding topics on the COSMOS message bus The problem can be simple event or complex series of events ii a solution which in its simplest form can be the URI of an loT service A solution can be primitive 1 task loT service or complex series of loT services One of the main disadvantages of CBR store compute trade offs because of large case bases is tackled as most of the VEs are going to have their own light weight case base stored locally As it mentioned in COSMOS WP5 D5 1 1 Section 4 1 4 we are going to develop an ontology based CBR planner and adopt initially the Flat Memory model In the use case that we are going to present in Section 5 the way the Cases are created is by using the method thermocase in the Monitor class of the DemoProject This method generates random numbers at fixed intervals for the values of temperature time etc Date 22 07 2014 Grant Agreement number 609043 Page 8 of 24 D5 2 1 Decentralized and Autonomous Things Management 2 2 2 Reasoner CBR is based on finding solutions to new problems by reusing older cases As a result the only component needed besides the CB in order to achieve CBR at its simplest form is a reasoner the Planner that can identify similarities between cases
27. ulateMaxTTL method of the Social Monitoring component we can take into account not only the initial recursiveness targeted but other criteria too like the number of friends Thus we can actually adjust this number if we deem it too high By initially using a simple mathematical function of a logarithmic nature we can adjust the ttl by using the friend count of the current VE as well as a target audience to be reached maxhits variable This variable will be extracted per domain and further code development and further experimentation with real data will refine the function used Finally if the experimental data points to such a possibility we will use an absolute upper limit for ttl in the sense of the six degrees of separation theory Following the calculation of the ttl from the previous method the getMaxTTL method will check whether the inputted ttl is above bounds and if so then adjust it accordingly Date 22 07 2014 Grant Agreement number 609043 Page 11 of 24 D5 2 1 Decentralized and Autonomous Things Management cosmos 4 Social Analysis 4 1 Description In order to support effective experience sharing and discovery social links establishment becomes necessary Sending a request only to VEs that are directly involved to its content is the most efficient way of communication as only VEs that can directly help or can offer the new required knowledge are informed This way the communication overhead is kept to a minim
28. um which is important for keeping the communication channel alive Thus the VEs need Friends and COSMOS has to develop a social environment that can support their discovery For this reason services like Friend Recommendation are developed Moreover the concepts of reliability trust and reputation are of outmost importance whenever a VE has to choose among cases offered as an answer to its request by different VEs For this reason services like Dependability calculation are provided too 4 2 Functionalities 4 2 1 Friends list and friend recommendation service During the registration process of a VE in the COSMOS platform it will be possible for the user to define pre existing friends inside the friend list of the localstore This is the most basic way a VE forms social bonds with others and such friends will have a number of benefits during the social monitoring or discovery mechanism phases e g greater priority Another way of acquiring friends will be through the friend discovery mechanism as explained in 3 2 3 However at any given point during the VE s lifecycle it can also petition the COSMOS Social Analysis component for a renewal of its friend list based on the new social metrics stored in the COSMOS Social Ontology for each VE Such a process is initiated by the VE acquiring a list of its friends through the retrieveFriends method After the list of friends is created the VE will also instantiate a List of weights to be use
29. valuate their performance upon request from any VE through the RebuildFriendList mechanism as is described in detail bellow The importance of maintaining such a file is mainly due to the fact that in principle socially dynamic behavior cannot be achieved without the platform being informed of changes in the social characteristics of VEs through the use of feedback mechanisms that will be developed in future prototypes At the moment all updates in the triple stores not counting changes for testing purposes are made through the use of the Jena libraries provided by the OntModel class and all queries retrievals are done using SPARQL Test changes are done directly to the owl files through Prot g Date 22 07 2014 Grant Agreement number 609043 Page 15 of 24 D5 2 1 Decentralized and Autonomous Things Management cosmos 7 Delivery and usage 7 1 Package information The deliverable contains the following projects and folders e Folders of Projects Developed o DemoProjectMaven o ExperienceGiverVEMaven o JennaPelletQuerryMaven o SocialAnalysisComponent o FriendOfFriend e Maven Configuration Files DemoProjectMaven pom xml ExperienceGiverVEMaven pom xml JennaPelletQuerryMaven pom xml SocialAnalysisComponent pom xml FriendOfFriend pom xml e Package Source Code location o DemoProjectMaven src o ExperienceGiverVEMaven src o JennaPelletQuerryMaven src O O O O OOO SocialAnalysisComponent src FriendOfFriend src
30. xes N2 File Name Title Date 22 07 2014 Grant Agreement number 609043 Page 2 of 24 D5 2 1 Decentralized and Autonomous Things Management cosmos Table of Contents Mable OP Contents DT A T 3 Table MS KH 4 L InioUB5HODn e de eoi petere Eid px edi esas a nu Dauid mue i reer paca ides 7 2 Planner Case Based eas Eos tte prre vu fre vnde gp soto ev tcu Macs pu C eEUA 8 2l Be 6 5 010 ener ere ES 8 241 F nctionalitieS scat sae ees odictor iius ciens Resa O 8 22 NS T IPAS E 8 2 2 24 Corn que ER 9 3 Decentralized Discovery Mecbanistiis uo niei mettre detto ts M Ern ind eae Rbtd 10 CAMERE eap clo C 10 3 2a JPunctonaltesu oriri i r a dte Edel Spic EM D Mui Uds 10 3 21 MC BSS CISCO UI NOE Tm 10 Sd JoT servi es ISCO VEL 5 cnnasccsesenasncnteaaeannatvoiwsinatececnadeciassaateetoeaacnersadnnesonne 11 3 2 3 Jens discovery osre einen E E ee es 11 p C l c lation of max ttln 11 4 Social unb siseses S 12 LAE rait 12 Zo UCI Oeste niece aiia A E E A tea inflat ees 12 4 2 Friends list and friend recommendation service sessssss 12 422 Dependability Index Calculatiogn ueei terrae tont tetris rera 13 5 DEMO Efficient Heating Service Luis been raten qve rx man euni nee n bobaa dus 14 5l London Scenario RR mm m m 1
31. y the same universally calculated and stored data type properties hasReliabilitylndex hasTrustIndex and hasReputationIndex the weights used in its calculation are most probably different for each VE The Social Analysis component retrieves the three mentioned Indexes and firstly calculates a normalized number by dividing trust and reputation This happens because by definition reputation is greater than trust Afterwards by using the weights provided by the VE it calculates the weighted sum of the normalized number and the reliability index in order to produce the new Dependability Index as a normalised number Date 22 07 2014 Grant Agreement number 609043 Page 13 of 24 D5 2 1 Decentralized and Autonomous Things Management cosmos 5 DEMO Efficient Heating Service The purpose of this section is to describe the use case that has been chosen in order to demonstrate the use of the main components and services of WP5 The different steps of the scenario are given below 5 1 London Scenario In smart home environment total energy consumption is measured in real time with the help of smart meters 1 Every flat is modelled as a VE Every flat contains a thermometer sensor and a boiler whose temperature can be measured sensor as well as set actuator 2 AVE registers to COSMOS and its loT services are defined and exposed through RESTful services Moreover COSMOS provides the VE with an initial set of friends
Download Pdf Manuals
Related Search