- The document analyzes whether existing Web of Things platforms truly follow the principles of the Web of Things.
- It identifies 12 key elements of the Web of Things, including device discovery, integration with the web, use of RESTful services, and security.
- It then evaluates 26 popular Web of Things platforms against these 12 elements. It finds that none of the platforms fully satisfy more than 10 of the 12 elements, and many platforms fail to properly address important elements like device discovery and security.
- The document concludes that while existing platforms provide some Web of Things capabilities, no platform currently offers a complete solution that covers all aspects of integrating physical devices with web clients and services as envisioned by the Web of
Hyperautomation and AI/ML: A Strategy for Digital Transformation Success.pdf
Do ”Web of Things Platforms” Truly Follow the Web of Things?
1. Do Web of Things Platforms Truly Follow
Web of Things?
Unit for Reasoning, Querying and Stream Processing
Insight Centre for Data Analytics,
National University of Ireland, Galway
Andreas Kamilaris, Ali Intizar
World Forum – IoT 2016
Reston, VA, USA- December 12-14, 2016
4. • Few of the emerging WoT
platforms
• Sorcades
• ThingWorx
• SpitFire
• Evrythng
• Open.Sen.se
• WoTKit
• Auto WoT
• Xively
Web of Things Platforms
5. • Our Motivation
• What are the common characteristics that identify the Web of
Things platforms?
• What are the must have functionalities to make a platform WoT
enabled?
• Do WoT platforms follow Web Standards?
• What is the status of the state of the art WoT platforms?
• How can we improve WoT platforms?
• How can we help to increase the speed of the evolution of WoT
platforms?
Analysis of Web of Things Platforms
6. • Our Methodology:
• Step 1: Identify the basic elements of the WoT, based on
the state of art literature and pioneering work in the field.
• Step 2: Locate the most popular platforms and frameworks
which claim to be ready for the WoT.
• Step 3: Analyze these platforms according to each of the
foundational WoT elements, as identified in Step 1.
• Step 4: Consider each of the basic elements of the WoT
one by one, examining how they have been
addressed/satisfied by those platforms.
Analysis of Web of Things Platforms
7. • Step 1: Identify the basic elements of the WoT
• Device Discovery:
• Locally (within the same physical space)
• Globally (through Web)
• Integration to the Web:
• Locally (within the same physical space)
• Globally (through Web)
Analysis of Web of Things Platforms
8. • Step 1: Identify the basic elements of the WoT
• Restful Services between the devices/Things
• Restful Services between the WoT platforms
Analysis of Web of Things Platforms
9. • Step 1: Identify the basic elements of the WoT
• Supported Data Formats
• Multiples Response Types
Analysis of Web of Things Platforms
10. • Step 1: Identify the basic elements of the WoT
• Security
• Authorization
• Authentication
• Message Encryption
Analysis of Web of Things Platforms
11. • Step 1: Identify the basic elements of the WoT
• Semantic Description
• Service Semantics: semantic description of services e.g. what they
provide and how to access them.
• Data Semantics: semantic description of data produced by the
devices e.g. what is data schema and how it is represented.
Analysis of Web of Things Platforms
12. • Step 1: Identify the basic elements of the WoT
• Mashup Services
• Create enhanced services by mash-up of the existing services
• Service re-usability and sharing
• Availability of services for re-use and share among multiple users
• Syndication Techniques/Web Messaging
• Create enhanced services by mash-up of the existing services
Analysis of Web of Things Platforms
13. • Step 2: Locate the most popular Web of Things Platforms
• Web of Things Platforms
• 26 most popular Web of Things Platforms
• Popularity is defined in terms of academic popularity e.g. more than
15 citations for the relevant research articles
• A combination of academic research proposal and commercial
products/startups
Analysis of Web of Things Platforms
14. • Step 2: Locate the most popular Web of Things Platforms
Analysis of Web of Things Platforms
No Platform/Project
1 TinyREST
2 pREST
3 SOCRADES
4 HomeWeb
5 SenseWeb
6 Sense2Web
7 Sensor.Network
8 SensorBase
9 Sensorpedia
10 Social Access Controller (SAC)
No Platform/Project
11 WoT framework for CPS
12 WebPlug
13 AutoWoT
14 SenseBox
15 Akribopoulos et al.
16 SemSense
17 VO Framework
18 WoTKit
19 GaaS
20 Dynamix
No Company
21 Xively
22 Evrythng
23 Open.Sen.se
24 ThingWorx
25 Paraimpu
26 SpitFire
15. • Step 3 & 4: Analyze and examine selected platforms to verify their
compliance to the common characteristics of WoT platforms
• Device Discovery
• The ”Achilles heel” of WoT
• The large majority of the platforms (21 out of 26) do not implement or adopt
any discovery protocol.
• From the few platforms dealing with discovery of devices, almost all focus on
local device discovery at the same physical space.
• TinyREST and Dynamix use UPnP
• SOCRADES uses WS-Discovery.
• HomeWeb [6] has implemented a lightweight equivalent of WS-Discovery,
more appropriate for the WoT.
• The only platform focusing on global discovery of devices on the web is
SpitFire, which employs an interesting discovery technique based on web
crawlers.
Analysis of Web of Things Platforms
16. • Step 3 & 4: Analyze and examine selected platforms to
verify their compliance to the common characteristics
of WoT platforms
• Integration to the Web
• Satisfied only by a subset of the platforms (10 out of 26).
• pREST, SOCRADES, Akribopoulos et al. and the VO Framework
employ embedded HTTP servers on SunSPOTs and Arduinos.
• HomeWeb and SpitFire use 6LoWPAN and CoAP.
• SenseWeb and SAC use gateways
• Evrythng developed its own embedded device toolkits for helping
developers build their own WoT applications.
Analysis of Web of Things Platforms
17. • Step 3 & 4: Analyze and examine selected platforms to
verify their compliance to the common characteristics of
WoT platforms
• Restful Services Devices
• 10 out of the 26 platforms suggest a RESTful interaction with WoT
devices.
• Gateways are the default option for constrained things such as RFID
tags, QR codes and barcodes.
• SenseWeb supports interaction through WS-* services.
• SOCRADES offers also DPWS-based support.
• The rest of 16 platforms offer only indirect ways to interact with
physical devices through RESTful interfaces.
Analysis of Web of Things Platforms
18. • Step 3 & 4: Analyze and examine selected platforms to
verify their compliance to the common characteristics
of WoT platforms
• Restful Services Platforms
• Most satisfied requirement by the platforms (18 out 26).
• Platforms provide RESTful APIs for users to interact with WoT
devices/services and data collected.
• SenseWeb and SensorBase offer interaction based on WS-*.
• The on-going battle REST Vs. WS-* .
Analysis of Web of Things Platforms
19. • Step 3 & 4: Analyze and examine selected platforms to
verify their compliance to the common characteristics of
WoT platforms
• Data Formats
• Most platforms employ XML, JSON, CSV and HTML.
• XML is the most widely used (17 out of 26 platforms).
• Atom (Sensorpedia, WebPlug and Xively), XHTML (SOCRADES)
• GeoRSS (Sensorpedia), KML (WoTKit),
• GeoML (WoT framework for CPS), EEML and RSS (Xively).
• RDF is employed in Sense2Web, the WoT framework for CPS and in
SpitFire.
• SenseWeb, Open.Sen.se and ThingWorx use only plain text!
Analysis of Web of Things Platforms
20. • Step 3 & 4: Analyze and examine selected platforms to
verify their compliance to the common characteristics
of WoT platforms
• Multiple Representation of Response Types
• A powerful element of the WoT is the opportunity to negotiate
among multiple representations of the HTTP responses from
things, their gateways or the platforms themselves.
• Negotiation should be based on various well-known data formats
on Web.
• From the 26 platforms, only 16 satisfy this important requirement.
Analysis of Web of Things Platforms
21. • Step 3 & 4: Analyze and examine selected platforms to verify their
compliance to the common characteristics of WoT platforms
• Security
• 14 platforms do not consider security at all.
• HomeWeb, SenseWeb, SensorBase and Open.Sen.se offer only basic user
authentication.
• Sensor.Network, Xively, Evrythng, ThingWorx and Paraimpu require an API
key on the HTTP header, to call RESTful services.
• Sensor.Network and Paraimpu support HTTP Secure (HTTPS),
• Sensorpedia supports OpenID.
• OAuth 2.0 is adopted by Sensorpedia, SAC and Evrythng.
• Xively is the platform with the most complete security profile, supporting
among others TLS and DTLS for privacy and data integrity during message
communication.
Analysis of Web of Things Platforms
22. • Step 3 & 4: Analyze and examine selected platforms to verify their
compliance to the common characteristics of WoT platforms
• Service Semantics
• More than half (14 out of 26) platforms provide semantics for interacting with
their APIs and devices.
• SOCRADES and GaaS adopt WSDL, HomeWeb proposes WADL.
• HomeWeb and SAC also use HTML links to navigate between services.
• AutoWoT adopts microformats.
• SenseBox prefers the OGC Sensor Web Enablement (SWE) standards
• SpitFire uses sensor meta-data to describe sensors’ services.
• Few research-based platforms (Sensor.Network, WoTKit), as well as all
commercial platforms (Xively, Evrythng, Open.Sen.se, ThingWorx and
Paraimpu) provide detailed API documentations or SDKs in various
programming languages.
Analysis of Web of Things Platforms
23. • Step 3 & 4: Analyze and examine selected platforms to verify their
compliance to the common characteristics of WoT platforms
• Data Semantics
• 16 out of 26 platforms do not provide any semantic descriptions of the data
exchanged between the things/platform and the user.
• SensorBase describes sensor data by SensorML, Akribopoulos et al. use
WiseML.
• Sensor.Network employs XML descriptors.
• The rest platforms (pREST, Sense2Web, SenseBox, SemSense, VO
Framework, Evrythng and SpitFire) employ Semantic Web technologies such
as RDF, OWL, SPARQL as well as the OGC Observations and Measurements
(O&M) standard.
• The commercial platforms do not employ semantics, with the exception of
Evrythng.
Analysis of Web of Things Platforms
24. • Step 3 & 4: Analyze and examine selected platforms to
verify their compliance to the common characteristics
of WoT platforms
• Mashups
• Fifteen platforms provide some mashup functionality (e.g. sensor
maps, plots, widgets)
• Only HomeWeb, WoTKit, GaaS, Evrythng, Open.Sen.se,
ThingWorx and Paraimpu offer mashup builders/composers.
• Mashup editors seem to be recognized by the commercial
initiatives as an integral feature provided to their end users.
Analysis of Web of Things Platforms
25. • Step 3 & 4: Analyze and examine selected platforms to
verify their compliance to the common characteristics of
WoT platforms
• Sharing and Re-usability
• Sharing of WoT devices and services among online contacts is a
practice not adopted by most platforms, except from HomeWeb, SAC
and Paraimpu.
• HomeWeb shares things and their services among Facebook friends.
• SAC delegates access to things based on Facebook and Twitter APIs
(to locate trusted contacts).
• Paraimpu shares things among other Paraimpu users.
Analysis of Web of Things Platforms
26. • Step 3 & 4: Analyze and examine selected platforms to verify their
compliance to the common characteristics of WoT platforms
• Syndication Techniques/ Web Messaging
• Fifteen platforms employ syndication and/or publish/subscribe messaging
techniques for interacting with things.
• Sensorpedia and SAC use ATOMPub as a syndication.
• TinyREST and pREST use dedicated subscription services on things.
• The concept of Web hooks is employed in HomeWeb, Sensor.Network,
SenseBox (push notifications), WebPlug, Open.Sen.se and Paraimpu.
• SOCRADES and GaaS select WS-Notification (from WS-*).
• VO Framework proposes the OSGi Eventing Service.
• Commercial platforms prefer more scalable publish/subscribe messaging
brokers such as MQTT (Xively and Evrythng) and XMPP.
Analysis of Web of Things Platforms
27. No Platform/Project
1 TinyREST
2 pREST
3 SOCRADES
4 HomeWeb
5 SenseWeb
6 Sense2Web
7 Sensor.Network
8 SensorBase
9 Sensorpedia
10 Social Access Controller (SAC)
No Platform/Project
11 WoT framework for CPS
12 WebPlug
13 AutoWoT
14 SenseBox
15 Akribopoulos et al.
16 SemSense
17 VO Framework
18 WoTKit
19 GaaS
20 Dynamix
No Company
21 Xively
22 Evrythng
23 Open.Sen.se
24 ThingWorx
25 Paraimpu
26 SpitFire
Supporting 2-3 Elements
Supporting 4-6 Elements
Supporting 7-9 Elements
Supporting 10 or
more Elements
Conclusion
28. • Global discovery on the WoT is still an open issue.
• Description of services is addressed mainly by the commercial approaches.
• Data semantics is supported mostly by the research initiatives, by means of
Semantic Web technologies.
• Evrythng is the most complete platform from the ones reviewed.
• None of the 26 platforms fully adopt more than 10 of the 12 basic elements
identified.
There do not exist any complete WoT platforms yet, covering the whole cycle
from physical things to web clients, addressing effectively all the issues of
discovery, interaction, service/data description, messaging, security, sharing
and mashup creation.
Conclusion
Editor's Notes
Evrythng and HomeWeb are the platforms closer to the concepts of the WoT, both scoring 10/12.
SemSense and Dynamix develop only two elements, TinyREST and Akribopoulos et al. support only three, while the WoT framework for CPS, Sense2Web, SensorBase, Sensorpedia, AutoWoT and WoTKit provide only five of the 12 elements.
Hence, we suggest that academia and enterprise should use the ”Web of Things” label more moderately, with some caution