Your SlideShare is downloading. ×
  • Like
Towards a comprehensive approach to spontaneous self-composition in pervasive ecosystems
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×

Now you can save presentations on your phone or tablet

Available for both IPhone and Android

Text the download link to your phone

Standard text messaging rates apply

Towards a comprehensive approach to spontaneous self-composition in pervasive ecosystems

  • 191 views
Published

Pervasive service ecosystems are emerging as a new paradigm for understanding and designing future pervasive computing systems featuring high degrees of scale, openness, adaptivity and toleration of …

Pervasive service ecosystems are emerging as a new paradigm for understanding and designing future pervasive computing systems featuring high degrees of scale, openness, adaptivity and toleration of long-term evolution. A key issue in this context is making certain patterns of behaviour emerge without any supervision or design-time intention, and a primary example is the fully-spontaneous composition of services, possibly at multiple levels. We argue that this can be successfully achieved only by a comprehensive approach exploiting together the main ingredients proposed so far in literature: (i) existence of intel- ligent components finding proper (semantic) matches of service descriptions, (ii) use of distributed evolutionary techniques to dynamically select appropriate ways of composing services, and (iii) approaches in which rating quality of composition is solely based on their successful exploitation. This proposal is presented through an example of spontaneous composition in crowd steering services.

  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Be the first to comment
    Be the first to like this
No Downloads

Views

Total Views
191
On SlideShare
0
From Embeds
0
Number of Embeds
0

Actions

Shares
Downloads
3
Comments
0
Likes
0

Embeds 0

No embeds

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
    No notes for slide

Transcript

  • 1. Towards a comprehensive approach to spontaneous self-composition in pervasive ecosystems Sara Montagna Mirko Viroli Danilo Pianini Jose Luis Fernandez-Marquez Email: sara.montagna@unibo.it ` A LMA M ATER S TUDIORUM—Universita di Bologna a Cesena University of Geneva, Switzerland Undicesimo Workshop Nazionale “Dagli Oggetti agli Agenti” (WOA’12) Milano-Bicocca, Italy, 17-19 September 2012Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 1 / 22
  • 2. 1 A comprehensive approach for pervasive ecosystems2 The self-composition issue in pervasive service ecosystems3 Gradient self-compositions4 Towards simulation of gradient self-compositions Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 2 / 22
  • 3. A comprehensive approach for pervasive ecosystemsOutline1 A comprehensive approach for pervasive ecosystems2 The self-composition issue in pervasive service ecosystems3 Gradient self-compositions4 Towards simulation of gradient self-compositions Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 3 / 22
  • 4. A comprehensive approach for pervasive ecosystemsPervasive service ecosystems [VPMS12]SAPERE Vision Mobile devices, people, software services, data, events Individuals Self-organisation enacted at the system level High degrees of scale openness adaptivity toleration of long-term evolution Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 4 / 22
  • 5. A comprehensive approach for pervasive ecosystemsAbstract Architecture Figure : An architectural view of a pervasive ecosystem. Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 5 / 22
  • 6. A comprehensive approach for pervasive ecosystemsLive semantic annotationsBasic block of semantic chemistry A unified description for every entity A unique LSA-id plus a semantic description (SD) RDF-inspired set of multi-valued properties Contains everything is needed for describing the entityExample: gradient source annotation:id314 mid:#loc :loc117; sos:type sos:source; sos:step "0"; sos:sourceid "341AB2" sos:aggr_prop sos:sourceid; sos:r_diff "10"; sos:r_ctx "100" Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 6 / 22
  • 7. A comprehensive approach for pervasive ecosystemsEco-LawsLanguage of semantic chemistry Chemical rules over LSA templates P+...+P --r--> Q+...+Q Constrained variables written ?V(filter) Check for presence “+”, absence “-” or unique existence “=” They can diffuse an LSA in the neighborhood They can aggregate LSAs like in chemical bondingExample: source pump?SOURCE sos:type sos:source; sos:aggr_prop ?P; sos:step ?T; sos:r_diff ?R--?R-->?SOURCE sos:step =(?T+1) + ?GRAD(?GRAD clones ?SOURCE)sos:type -sos:source sos:diff sos:aggr; sos:dist "0"; sos:orientation "here" Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 7 / 22
  • 8. The self-composition issue in pervasive service ecosystemsOutline1 A comprehensive approach for pervasive ecosystems2 The self-composition issue in pervasive service ecosystems3 Gradient self-compositions4 Towards simulation of gradient self-compositions Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 8 / 22
  • 9. The self-composition issue in pervasive service ecosystemsSelf-CompositionKey issue Patterns of behaviour emerge without any supervision Example: fully-spontaneous composition of services, possibly at multiple levels Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 9 / 22
  • 10. The self-composition issue in pervasive service ecosystemsSome self-composition issues Composition of services not explicitly designed to coordinate Composition of “compatible” services Creation of “meaningful” services Context awareness Multi-level composition Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 10 / 22
  • 11. The self-composition issue in pervasive service ecosystemsSelf-composition in service ecosystemsComposition of services in literature 1 Service Composition in SOA – advanced semantic matching 2 Evolutionary techniques 3 Competition-based approachesAll the above, altogether 1 Choice of the services to compose 2 Pre-selection of “promising” compositions 3 Fine parameter tuning 4 Service evaluation metrics 5 Best services must be promoted Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 11 / 22
  • 12. Gradient self-compositionsOutline1 A comprehensive approach for pervasive ecosystems2 The self-composition issue in pervasive service ecosystems3 Gradient self-compositions4 Towards simulation of gradient self-compositions Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 12 / 22
  • 13. Gradient self-compositionsParadigmatic Example: Crowd SteeringGoal and requirements Guide people towards POIs POIs chosen with respect to people’s interests Avoiding obstacles (incl. crowds) no supervisionScenario A museum with a dense network of sensor nodes Sensing of the presence of nearby visitors Computation abilities Visitors own smartphone devices holding their preferencesServices available Gradient service Those provided by sensors (e.g., crowd detection service) Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 13 / 22
  • 14. Gradient self-compositionsA Prototype Solution for Gradient CompositionComposition “composition recommender” agents computing all the available compositionsContextualisation Gradients are contextualisedFeedback Users public their “satisfaction” once they used the serviceChoice Users tend to prefer lower distance and higher satisfactionEvaporation Satisfaction fades with timeEvolution Parameters tuning by agents using evolutionary techniques Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 14 / 22
  • 15. Gradient self-compositionsEco-Laws for Gradient[PUMP]: An annotation of type source continuously injects the initial gradient annotation?SOURCE sos:type sos:source; sos:aggr_prop ?P; sos:step ?T; sos:r_diff ?R; sos:r_ctx ?RC--?R-->?SOURCE sos:step =(?T+1) +?GRAD(?GRAD clones ?SOURCE) sos:type -sos:source sos:diff sos:aggr; sos:dist "0"; sos:orientation "here"[DIFF] A gradient annotation is cloned in a neighbour, with distance increased and updated orientation?GRAD sos:type sos:diff; sos:dist ?D; sos:r_diff ?R +?NEIGH mid:type mid:#neigh; mid:remote ?L; mid:orientation ?O; mid:distance ?D2--?R-->?GRAD + ?NEIGH +?GRAD1(?GRAD1 clones ?GRAD) sos:type -sos:diff sos:ctx; sos:dist =(?D+?D2); sos:orient =?O; mid:#loc ?L[CTX] A contextualising annotation is transformed back into an annotation to be diffused?GRAD sos:type sos:ctx; sos:r_ctx ?RC --?RC-> ?GRAD sos:type sos:-ctx sos:diff;[YOUNGEST] Of two annotations the one with newest information is kept?ANN1 sos:type sos:aggr; sos:aggr_prop ?P; ?P =[?C]; sos:step ?T1 +?ANN2 sos:type sos:aggr; sos:aggr_prop ?P2; ?P2 =[?C]; sos:step ?T2(?T2<?T1)--->?ANN1[SHORTEST] Of two annotations the one with shortest distance from source is kept?ANN1 sos:type sos:aggr; sos:aggr_prop ?P; ?P =[?C]; sos:dist ?D1; sos:step ?T +?ANN2 sos:type sos:aggr; sos:aggr_prop ?P2; ?P2 =[?C]; sos:dist ?D2(?D2>=?D1); sos:step ?T--->?ANN1[DECAY] An annotation decays?GRA sos:type sos:diff; sos:r_dec ?RD --?RD-> 0 Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 15 / 22
  • 16. Gradient self-compositionsEco-Laws for Gradient Composition[COMPOSITION] The gradient source is composed with the crowd service?SOURCE sos:type sos:source; scm:satisfaction ?S + ?CROWD scm:type crowd; crowd:level ?CL--->?SOURCE + ?CSOURCE(?CSOURCE clones ?SOURCE) scm:property sos:dist;scm:parameters scm:crowd_op ?CF; scm:crowd_op ?CF*?CL[CONTEXTUALISATION] If sensors perceive crowd, the gradient distance is augmented?GRAD sos:type sos ctx; sos:dist ?D; scm:property sos:dist;scm:parameters scm:crowd_op scm:crowd_factor; scm:crowd_factor ?CF; scm:crowd_op ?CF*?CL +?CROWD scm:type crowd; crowd:level ?CL--->?CROWD + ?GRAD sos:type -sos:ctx sos:diff; sos:dist =(?D+?CF*?CL)[FEEDBACK] Feedbacks are used to update the satisfaction values?FEEDBACK scm:parameters scm:crowd_op; scm:feedback scm:velocity; scm:velocity ?V +?GRAD scm:satisfaction ?S; scm:parameters scm:crowd_op--->?GRAD scm:satisfaction =(?S+?V)[EVAPORATION] The gradient satisfaction value gets decreased?GRAD scm:satisfaction ?S; scm:factor_ev ?FE; scm:r_ev ?RE--?RE->?GRAD scm:satisfaction =(?FE*?S)[DECAY] If the gradient satisfaction value becomes zero that composition is removed?GRAD scm:satisfaction "0";---> Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 16 / 22
  • 17. Towards simulation of gradient self-compositionsOutline1 A comprehensive approach for pervasive ecosystems2 The self-composition issue in pervasive service ecosystems3 Gradient self-compositions4 Towards simulation of gradient self-compositions Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 17 / 22
  • 18. Towards simulation of gradient self-compositionsSimulation as a proof of concepts Conducted using A LCHEMIST [PMV11] Early experiments on gradient composition with crowd level Different compositions with different crowd relevance different composite gradients Satisfaction value measures the time to POI Users choose one gradient considering distance and satisfaction Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 18 / 22
  • 19. Towards simulation of gradient self-compositionsSimulation Results I Figure : Satisfaction values for different compositions changing over time. Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 19 / 22
  • 20. Towards simulation of gradient self-compositionsSimulation Results II Figure : Satisfaction values for different compositions changing over time. Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 20 / 22
  • 21. ReferencesReferences I Danilo Pianini, Sara Montagna, and Mirko Viroli. A chemical inspired simulation framework for pervasive services ecosystems. In Proceedings of the Federated Conference on Computer Science and Information Systems, pages 675–682. IEEE Computer Society Press, 2011. Mirko Viroli, Danilo Pianini, Sara Montagna, and Graeme Stevenson. Pervasive ecosystems: a coordination model based on semantic chemistry. In 27th Annual ACM Symposium on Applied Computing (SAC 2012), pages 295–302. ACM, 2012. Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 21 / 22
  • 22. ReferencesTowards a comprehensive approach to spontaneous self-composition in pervasive ecosystems Sara Montagna Mirko Viroli Danilo Pianini Jose Luis Fernandez-Marquez Email: sara.montagna@unibo.it ` A LMA M ATER S TUDIORUM—Universita di Bologna a Cesena University of Geneva, Switzerland Undicesimo Workshop Nazionale “Dagli Oggetti agli Agenti” (WOA’12) Milano-Bicocca, Italy, 17-19 September 2012Montagna et al. (UniBo/UniGe) Self-composition of services WOA’12 22 / 22