Tractable Policy Management Framework DAIS ITA for IoT Geeth de - PowerPoint PPT Presentation

Tractable Policy Management Framework DAIS ITA for IoT Geeth de Mel, IBM United Kingdom Ltd., UK Distributed Analytics Emre Goynugur, Murat Sensoy, Ozyegin University, Turkey and Information Seraphin Calo, IBM Thomas J. Watson Research Center, USA Science Geeth de Mel, IBM UK International Technology Alliance SPIE DSS 2017 – Conference 10190 Ground/Air Multisensor Interoperability, Integration, and Networking for dais-ita.org Persistent ISR VIII

Overview: What is IoT?

Explosion in IoT ? 3 1 http://www.forbes.com/sites/bernardmarr/2015/09/30/big-data-20-mind-boggling-facts-everyone-must-read/#425a31276c1d 2 http://www.whizpr.be/upload/medialab/21/company/Media_Presentation_2012_DigiUniverseFINAL1.pdf 3 https://www.ibm.com/internet-of-things/

Adaptive Interaction Governance for IoT • Value added is from the transformations done by services CONSTRAINTS • Why Policies • Features of an IoT Policy Framework • End users want to program their own devices (e.g. IFTTT) • Expressiveness • Each user has different preferences • Efficient Reasoning over a large amount of data • Policies as soft constraints • Automatic conflict detection • It is not trivial to learn some rules through user behaviors • Automatic conflict resolution • it is forbidden to play music in room X. why? • Learning takes time, rules are instant

Traditional Policy Frameworks: Issues w.r.t IoT Policy Creation User • Bulky Tool • Designed for IT systems • Verbose Policy Management Policy Tool • Human oriented specification Repository • Lack of transparency Policy Decision • Explicit rules for each situation Point • Rigid Policy Audit Tool • E.g., Pace maker Policy Enforcement Point • Responsibility issues System • E.g., Driverless car Logs • Policy Refinement Requester

A new Paradigm Adaptive Interactive Iterative & Contextual Stateful 6

Tractable Policy Management Framework for IoT Prohibitions • A semantic policy language based on OWL-QL χ : ρ ? d : Device (? d ) N P Baby (? b ) ∧ Sleeping (? b ) ∧ inFlat (? b , ? f ) ∧ inFlat (? d , ? f ) • Efficiency, expressiveness, and conflict α a : ϕ ? a : MakeSound (? a ) detection e Awake (? b ) c 10 . 0 • Model devices as a set of services Obligations • Television: an Internet browser, a display, a speaker and so forth. χ : ρ ? d : Doorbell (? x ) N O SomeoneAtDoor (? e ) ∧ producedBy (? e , ? x ) ∧ • Define high-level policies w.r.t. ontology α belongsToFlat (? x , ? f ) ∧ hasResident (? f , ? p ) ∧ Adult (? p ) a : ϕ ? a : NotifyWithSound (? a ) ∧ hasTarget (? a , ? p ) and refine them to device/service level e c 4 . 0 • Use an AI planner to minimize the violation costs

What is an Ontology? “formal, explicit specification of a shared conceptualisation” - T. Gruber (1993)

Tractable Policy Management Framework for IoT Prohibitions • A semantic policy language based on OWL-QL χ : ρ ? d : Device (? d ) N P Baby (? b ) ∧ Sleeping (? b ) ∧ inFlat (? b , ? f ) ∧ inFlat (? d , ? f ) • Efficiency, expressiveness, and conflict α a : ϕ ? a : MakeSound (? a ) detection e Awake (? b ) c 10 . 0 • Model devices as a set of services Obligations • Television: an Internet browser, a display, a speaker and so forth. χ : ρ ? d : Doorbell (? x ) N O SomeoneAtDoor (? e ) ∧ producedBy (? e , ? x ) ∧ • Define high-level policies w.r.t. ontology α belongsToFlat (? x , ? f ) ∧ hasResident (? f , ? p ) ∧ Adult (? p ) a : ϕ ? a : NotifyWithSound (? a ) ∧ hasTarget (? a , ? p ) and refine them to device/service level e c 4 . 0 • Use an AI planner to minimize the violation costs

Reasoning about Policies for IoT Policy Conflicts Architecture • Two policies conflict if all three conditions hold true: • Both policies are refined to the same addressee • One is an obligation and the other is a prohibition • Activation conditions hold true in a consistent world state • Conflict detection is a query containment problem • Apply query freezing to reduce it to query answering

Policy Conflict Resolution Planning • Suitable to integrate with a knowledge base • known information is our domain • current state is the initial state • Convert device services into domain actions • Embed ontology information into the domain file • e.g. reasoning, consistency check • Violated policy costs are added to the plan cost • Preference 1 - and Cost 2 - based Planning Find alternative ways to fulfill an obligation

Conclusions • IoT systems should be autonomous and intelligent • Need a scalable, efficient, and expressive policy framework • E.g., OWL-QL knowledge base • Devices should collaborate (also with web services) • Automated detection and resolution of conflicts is essential • Reformulate conflict resolution as a planning problem • Future Work • Violation costs may be learned from user history • Stream data and its impact on planning 12

Any Questions?

Acknowledgement This research was sponsored by the U.S. Army Research Laboratory and the U.K. Ministry of Defence under Agreement Number W911NF-16-3-0001. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the U.S. Army Research Laboratory, the U.S. Government, the U.K. Ministry of Defence or the U.K. Government. The U.S. and U.K. Governments are authorized to reproduce and distribute reprints for Government purposes notwithstanding any copy-right notation hereon. Dr. Şensoy thanks the U.S. Army Research Laboratory for its support under grant W911NF-14-1-0199 and The Scientific and Technological Research Council of Turkey (TUBITAK) for its support under grant 113E238.