View source for OntologySummit2015 Related Standards and Synergies for Emerging IoT Ontologies Synthesis

'''Related Standards and Synergies for Emerging IoT Ontologies'''

[[Category:OntologySummit]]
[[Category:OntologySummit2015]]

'''Goal''' 
Identify relevant or de facto standards involved in the adoption of ontologies for the internet of things.<br><br>
A goal of Track D was to examine the work of IoT standards development organization (SDOs) and to study the role of standards in current implementations.  Insight into current standards practices in IoT could guide prospective ontology implementations.<br><br>
'''Standards Commentary'''
The yin and yang of standards affects designers and products in unpredictable ways. Standards that have outlived their usefulness (the B-school use case is the buggy whip) lie at one extreme. At the other end are creative, essential human needs to innovate without the baggage of decoding the language and potential bloat of a software standard. <br>
<br>A decade-old example that predated IoT's entry into common parlance was Project Drishti (Ran, Helal, & Moore, 2004)   The investigators sought to integrate data streams from RFID tags, GPS and wireless networks to aid the visually impaired in common navigation tasks. There were numerous other ideas in the wearable and ubiquitous computing literature and elsewhere, even in science fiction. <br><br>
Fast forward a decade and the number of data sources has multiplied. Big Data is competing with IoT for attention, and legitimately so, as noted in the 2014 Ontology Summit. This has created terrific momentum, especially for Big Data and the Apache stack which owns most of the developer mindshare about this paradigm shift. A convergence of open source projects, cloud computing and a steady march toward web-enabled applications has facilitated big data, but has the same occurred for IoT? There does not seem to be an IoT equivalent for the shift represented by the Apache stack with Hadoop at its center.<br><br>
Perhaps. Track D has not fully examined existing SDOs and projects to make a fair assessment. That said, it seems clear that there are many efforts underway, and that full coordination (compliance?) with standards or SDOs is not a prerequisite for building a workable system. Standards benefits in IoT as might accrue from incorporating ontologies may be more evident as systems mature than at this early stage of IoT work. A complex system requiring many different human and organizational roles, processing speeds and volume might need an ontology as its associated sensor grid shifts beneath it. Consider sensor devices being upgraded at the same pace as today's smartphones. <br><br>
'''IoT SDO Speakers''' 
''IEEE 21451-1-4''
William Miller (MaCT-USA), chair of ISO/IEC/IEEE 21451-1-4 <br>Presented the work of the IEEE P21451-1-4 SDO. This SDO is responsible for developing the XMPP standard. His IoT standards survey called out REST, Constrained Application Protocol (CoAP) as well as MQTT as influential. <br><br>
The standard's full name is "XMPP Interface for Smart Transducers and the 1st Semantic Web 3.0 Standard for IoT." Smart transducers originate modestly rich semantic information which the standard seeks to canonize. These and similar devices can be network-enabled, can summarize events, manage time windows, allow for discovery, sometimes paired by Transducer Electronic Data Sheets (TEDS) which  perform these duties for less capable devices. XMPP protocols have SOA, pub/sub, support discovery and measures to protect signals and authenticate devices. <br><br>
Is this different from any instrumentation scenario, such as medical or industrial automation in the 70's? Yes and no. The Implementation challenges were similar, but the computation pipeline was slower, involved fewer elements, and the stack not nearly so tall.<br><br>

'''MQTT''' <br>
Geoff Brown, Oasis Message Queuing Telemetry Transport (MQTT) Chair<br> provided a glimpse into that IoT standard. <br><br>
'''Other Standards'''
Legacy standards may represent design pattern histories. Established IoT standards include Distributed Control Systems (DCS) and Supervisory Control and Data Acquisition (SCADA).<br>
A collection of IoT SDOs and initiatives was presented by Underwood (slide 11). <br>
Postscapes lists Protocols and Standards [http://postscapes.com/internet-of-things-protocols] as well as selected IoT projects [http://postscapes.com/internet-of-things-software-guide], which include embedded operating systems.<br>
Due to their ubiquity and market volume, smartphone operating systems should be considered as potential IoT sources, brokers, aggregators, transducers and/or analytics resources.<br><br>
'''IoT Ontology Success Stories'''<br>
The W3C Semantic Sensor Network Ontology (OWL 2) and the OGC Sensor Web Enablement project (including SensorML, a Transducer Model Language, a Sensor Observations Service, Sensor Planning Service) efforts were cited by speaker Henson (Bosch).   Henson also cited the OGC Spatial Data on the Web planned for 2016. <br>
The GraphOfThings project incorporates SPARQL and the Continuous Query Evaluation over Linked Stream (CQELS) tool. Intellego leverages OWL, RDF and the SSN Ontology. <br>
The Track B presentation by Hodges cited highly curated biomedical ontologies. <br><br>
'''Software Design Choice and Opportunities''' <br>
Standards seen by developers as overly complex may be ignored in favor of ad hoc solutions that are -- or are seen as -- more flexible, easier to implement or lower cost. Except in a few highly regulated industries, software standards are omnipresent, but often not those promulgated by SDOs. Windows and Microsoft Office set down multi-faceted standards footprints that continue to influence designs, but they were not templates set forth by SDOs. <br>
Standards which govern access to data may represent significantly different sets of design choices for developers. Just as browser standards influenced UI and REST, the need to access cloud data resources at Facebook, Twitter and [[LinkedIn]] has fostered APIs that have been adopted by countless web sites and applications. To date, the Linked Data movement has continued to evolve but has not become a major influencer for applications development. While it is not the result of a conscientious app survey, <br><br>
'''Gaps, Loose Ends, Hopeful Signs'''<br>
'''Signs'''<br>
<i>Ongoing Standards Surveillance</i>  We have not fully explored all the facets of standards influencing. (Refer to "Standards Influence Maze," (Jakobs, K., cited on Underwood slide 6).<br> 
Are we seeing uptake of ontologies in Github for IoT systems building? (A measurable indicator.)<br>
Can an IoT equivalent to Google Search identify the scope of available end points for some domains?<br>
Are sensor manufacturers making design choices that can affect the perceived need for an ontological solution, or is this essentially a system-building undertaking?<br><br>
'''Loose Ends, Gaps and Forecasts'''<br>
Mainstreaming Wider deployments in process control and factory automation settings Standards have been in place in these settings for decades. Merging ontology solutions for into established problem subsets seems feasible and will be a signal that IoT ontologies have advanced closer to the mainstream.<br><br>
''Ontology Embedding'' The increased use of smart devices, store-and-forward, embedded intelligence automated data fusion (perhaps especially for geospatial aspects) suggests that ontology embedding could become a design pattern. The pattern could be used in building intelligent IoT, but ontology embedding within sensor systems themselves is possible. Metadata for discovery and provenance from devices are possible starting points.<br><br>
''Exploitation of (Lazy) Developer Pain Points'' Known problem areas in IoT exist across many different types of sensors. These include security, privacy, signal noise, reliability, configuration management, infrastructure dependency and other known architectural nuisances. Standard solutions in any one of these areas could catch on because it solves a well-defined problem that is tangential to an architect or sponsor's main system objectives. <br><br>
''Specialized Engines'' Reusable, high-complexity solutions might take hold to implement mathematical solutions in certain spaces, such as Gr&uuml;ninger's work with PDL in ERP or Spencer Breiner's category theory. (Related: Cloud Impact.)<br><br>
''Cloud Impact'' Because cloud engines such as Watson (Track X) will provide complex building blocks for architects, the challenge may be taken up by small groups or even sole developers working in green field problem spaces.<br><br>
''Fun Hardware Syndrome'' Sometimes collateral innovations co-occur with fun hardware developments. The smart car, or low cost commercial unmanned vehicles could spur ontology-rich solutions. The reasons for such developments are connected both to standards and to the attitudes (plus and minus) about existing standards. <br><br>
''Future IDE Innovation'' Will IoT need its own integrated development environment? IoT Test and development beds for IoT will likely require new combinations of devices, simulations, test data, standards, scalability exercises and more. If so, having integration ontology-building and testing built into an IDE through plugins and extensions could become an important, standards-fostering feature.<br><br>
'''Related Threads''' <br>
These are areas where related standards have evolved that could further inform the development of IoT ontologies. (Needs links to other track presentations, previous Summits.)
* Modsim (e.g., Yang Song, et al. 2012)
* Multisensor Correlation | Sensor Fusion Did these implementations use ontologies? If not, what design patterns were followed that led to alternative implementations?)
* [[DoD]] / DHS Situation Awareness
* Augmented reality
* Software Defined Networks (SDN)
* Retrospective: Related Lessons from Ontology Big Data 2014
* Domain-specific development frameworks 
* SDLC evolutionary history: middleware, intelligent agents, CEP. Ontology as middleware.