Countering Fragmentation With The Web Of Things - W3

Countering Fragmentationwith the Web of ThingsInteroperability across IoT platformsDave RaggettW3C27 April 2016

The Internet of Things – Bridging the SilosStill very immature, but with massive potentialLack of interoperability at the application levelMany platforms and associated standardsllAddressing broad range of different requirementsEnd to end security challenging across platformsFragmentation and Silos are holding back the potentialOpen or closed system?lllClosed systems incentive: controlOpen systems prompt: reduced costs and increased market sizeNeed for wide adoption of shared open standards2/35

IoT Landscape3/35

4The IoT Standardisation Challenge4/35

World Wide Web ConsortiumMission: lead the Web to its full potentiallThe Web is the world's largest vendor-neutral distributed application platformFounded by Sir Tim Berners-Lee, inventor of the Webll400 MembersMember-funded international organisationDevelops standards for Web and semantic technologieslllllHTML, CSS, scripting APIs, XML, SVG, VoiceXML,Semantic Web and Linked Data etc.Developer oriented, enabling cooperation betweenorganisations with very different backgroundsW3C patent policy for royalty free standardsW3C staff of engineers actively participating in standardisationIncreasingly involved in verticals: Mobile, TV, Automotive, Digital publishing5/35

6What we want to avoid 6/35

Web of Things – Inter-Platform standards for interoperabilityThe Web will enable a transition from costly monolithic software to open markets of apps7/35

Analogy with early days of networkingBefore the Internet, there were many noninteroperable network technologieslllIP made it simple to interconnect networks and createinteroperable services independent of the network technologiesThe Internet grew exponentially as the opportunities were realisedLikewise for the Web which took over from isolated informationservicesDirect analogy with today’s IoT silosand their lack of interoperabilityllThe Web of Things is the equivalent of IP for semanticinteroperability and end to end securityThe Web of Things will enable explosive growth as the barriers tointeroperability are torn down8/35

Web of ThingsApplications act on software objects thatstand for thingsllLocal “things”Remote “thingsRich descriptions for every “thing”llData models, semantics, metadataOntologies that describe “things”Things don’t need to be connectedlAbstract entities and unconnectedphysical objects9/35

Distributed Web of ThingslThing descriptions can beused to create proxies fora thing, allowing scriptsto interact with a localproxy for a remote entitylScripts can run on serversor as part of Web pages inWeb browser for humanmachine interfacelThing topologieslPeer to Peer, Peer to Peervia Cloud, Star, Device toCloud, Star to Cloud10/35

Communications Stack – Clean separation of concernsApplicationDeveloperApplicationScripts that define thing behaviour in terms of their properties, actionsand events, using APIs for control of sensor and actuator hardwareThingsSoftware objects that hold their stateAbstract thing to thing messagesSemantics and Metadata, Data models and DataTransferBindings of abstract messages to mechanisms provided by eachprotocol, including choice of communication pattern, e.g. pull, push,pub-sub, peer to peer, etc.TransportREST based protocols, e.g. HTTP, CoAPPub-Sub protocols, e.g. MQTT, XMPPOthers, including non IP transports, e.g. BluetoothNetworkUnderlying communication technology with support for exchange ofsimple messages (packets)Many technologies designed for different requirements(WoT focus)PlatformDeveloper(IoT focus)11/35

Scalability12/35

Web of Things for the Maker CommunityOpen source projects are underway, e.g.for the Arduino and more powerful MCUsArduino Ethernet Shieldl16 KB RAMlMicroSD card slotControlled through SPI buslPolling or H/W interruptlCost: 4.75 GBP on eBaylArduino Uno with ATmega328P MCUl2 KB RAM1 KB EEPROMl32 KB FLASHlllLots of I/O pinsCost: 2.33 GBP on eBayhttps://github.com/w3c/wot-arduino13/35

The Web of Things in the HomeBrowser for HMIAmbient or batteryoperated IoT devicesCloud 14/35

What kinds of metadata do we need for this?lThing lifecycles, data and interaction modelsllHow to interoperate with another platform?llMapping from thing descriptions to platform specific protocolsl IP address and port for IP based protocolsl Paths for REST based protocols such as CoAP & HTTPWhat communication patterns to use?llllllAs exposed to the applicationsPush, pull, pub-sub, peer to peerReal-time requirementsTransactional robustness & rollbacksMultiplexing and bufferingSleepy ambient & battery powered devicesSemantic models of things and their constraints15/35

16Data & Interaction ModelsMust be rich enough to cover broad range of use cases andplatformsllllProperties, actions and events carry valuesl Actions are asynchronous and can be passed a value,and may return a sequence of valuesValues as basic typesl Null, true, false, numbers, stringsCompound valuesl Arraysl Sets of name/value pairsl Thingsl StreamslIntegrity constraintsl On single values, e.g.l min/max, integer/floatl Across multiple valuesl Cardinality constraintsl Need for path expressionsComplicationsl Proxy chainsl Early and late bindingl Partially defined typesl Cyclic dependencies across thingsl Software dependenciesl Metadata constraints (versioning)16/35

SemanticslNeeded to ensure that platforms share the same meaning for the data they exchangelSimple approach is to define semantics as part of the system specificationsllBut this makes it easy to lose track when data is stored and passed to other systems, or whena system evolves to address changing requirementsBetter approach is to tag data as belonging to an ontology that describes therelationships between concepts in a machine interpretable wayllWhat kind of a thing is it?l e.g. a temperature sensorWhat are the domain constraints?l temperature sensors must describe their physical units, which must be from the set {Kelvin, Celsius, Fahrenheit}l Other ontologies could describe the location of the sensor and what it is measuring17/35

SemanticslOntologies allow information to be exchanged meaningfully in away that is independent of the data formats used for itstransmissionlOntologies further allow for checks that the information isconsistent with the domain modelslThis can cover richer constraints, e.g. temporal constraints acrossactions and propertieslW3C has a suite of standards for the Semantic Web and LinkedDatallRDF, XML, SPARQL, RDF-S, OWL, RIF, JSON-LD, RDF in CSV, . . .Enable semantic based search and composition of serviceslEnsure that compositions will use interoperable services18/35

Open QuestionslWhat needs to be standardised and how?lWhat are best practices and how can their use be encouraged?lW3C is in a good position to standardise cross domain ontologieslIndustry specific groups need to standardise domain specific ontologieslDifferent contexts place different requirements on domain modelslIdeally the ontologies are modularised to allow for easy extensionlFurther challenges from divergence between weakly coupledcommunities19/35

IWhat can we learn from the Web?lTim Berners-Lee, inventor of the Web, was keen for websites to add semantic annotationslTim’s original hopes for web sites to mark up their data failed to get tractionlMicroformats, a widely talked about alternative to RDFa, also failed to get tractionlWeb developers weren’t getting enough benefits for the effort they investedlSearch engine vendors to the rescue!llSimple flat semantics documented on schema.orgInstant benefit via how your website is presented in web search resultsl showing a restaurant on Google maps, along with the opening times20/35

Implications for the Web of ThingslWe need a way to search for services based upon their names, human readable descriptions and machineinterpretable semantic descriptionslWe need a way to compose services with the assurance that a given composition will work as expected. Thisimplies the need to check for semantic interoperabilitylWe need to look after security* and privacy.lllEnd to end security necessitates shared assumptions in respect to trust models, otherwise, platforms will only be able toshare data that is marked as publicly accessiblePrivacy and confidentiality are about the agreements between consenting parties.This relates to privacy preferences, privacy policies, sticky policies that remain associated with data as it is passed through thenetwork, the need to track provenance to ensure that data is handled in accordance with the agreement with the data owner,Service level agreements, Machine readable terms & conditions, Payments, and Automated negotiation.* In a general sense, including integrity, safety and resilience21/35

22Web of ThingsActivity22/35

Where Next? W3C Web of Things ActivitylW3C is chartering a Web of Things Working Group to standardise horizontal metadata vocabulariesllThis group is expected to launch later this yearW3C Web of Things Interest Group is re-charteringlllExpected to boost its work on reaching out to industry alliances and SDOsl Including Industrie 4.0Interoperability tests across platforms using open source implementationsFurther joint papers planned on security, privacy and requirements for open markets of services23/35

W3C Web of Things Interest GroupWorkshop in Berlin (June 2014)lllLaunch of Web of Things IG in 2015Chaired by Jörg Heuer, SiemensTask forcesThing descriptionsAPIs and protocolsDiscovery and provisioningSecurity, privacy and resilienceCommunications and collaborationStrong emphasis on implementation experiencellDemos and plug-festsHelps to build a shared understandingMontreal Face to Face, 11-13 April 201624/35

Members of the Web of Things Interest Group25/35

Joint White Paper on Semantic InteroperabilitylContributors are individuals from a range of industry alliances and standards development organisationsllllEditors from W3C, oneM2M, IEEE P2413Inspiration from many of the papers on semantic interoperabilityIllustrate the concept in terms of some use cases, along with the requirements, and views on how to enable semanticinteroperability within and across IoT platformsDiscuss best practices for ontology design and distinguish cross domain (horizontal) metadata and domain specific (vertical)metadatalOnce we have a stable version we will invite wider expert review and update the document accordinglylWe will seek broad dissemination of the document and will publish under a Creative Commons LicenselWe hope that the development process will help shape a common perspective across contributors and that the white paper willinfluence the agendas of working groups across the industry26/35

Liaisons and CollaborationsReaching out to industry alliances and SDO’s to driveconvergence to unleash the potentiallllllllPlattform Industrie 4.0Especially the “semantics” subgroupIndustrial Internet ConsortiumOpen Connectivity FoundationOPC FoundationIETF/IRTFoneM2MAIOTIAIOTI27/35

Enabling Vertical and Horizontal Integration forRealising the Potential for Digital AutomationW3C and OPC collaborating on enablingintegration across platforms throughmetadata standards for semanticinteroperability and end to end securityBusiness Levelhigh levelsof abstractionWebintegration alongofthe supply chainThingsintegration alongthe value chain*low levelsof abstractionField LevelThings Industrie 4.0 ComponentsIndustrie 4.0 Assets*value chain – the process oractivities by which a companyadds value to an article,including design, production,marketing, and the provision ofafter sales service28/35