6. Preparedfor:
SystemsResearchChallengesintheInternetofThings
Newcastle,Jan.2016
Evolution of the Internet (according to Google)
information graph connection to content
social graph connections amongst people
physical graph connections amongst things
TheInternetOfThingscreatesthephysicalgraph->that
changeshowweinteractwithobjectsandenvironments
Totalaccessandubiquityof
content
Adaptive,selfregulating
environmentsthatunderstand
contextandadjustaccordingly
TheInternetcreatedtheinformationgraph->thatchanged
howweproduce,access,shareandgenerateknowledge
Socialmediacreatedthesocialgraph->thatchangedhow
weestablishandfosterrelationshipwithothers
Enablingpowertothecrowd
3)Findtherightrelevantobjectsandconnections
Source: IEEE Internet of Things Vint Cerf, Google - December 15th 2015
9. Preparedfor:
SystemsResearchChallengesintheInternetofThings
Newcastle,Jan.2016
The connected washing machine example
1) Re-imagine statics object with the power of the web
W ashing Machine
Single purposed
No user customization
Rudimentary notification system
Not aware of its energy
consumption
Net
Context
Learning
Access to services and API’s
Access to other devices
+ + =
W ashing Machine
Adaptive
Efficient
Optimized
Personalized
Connectivity
New features
New services
New business models
web
1) Re-imagine isolated objects with the power of the net
Understanding the implications of what happens when an
‘ordinary’ object is connected to the net
Source: IEEE Internet of Things Vint Cerf, Google - December 15th 2015
11. Preparedfor:
SystemsResearchChallengesintheInternetofThings
Newcastle,Jan.2016
IoT ∩ Science
Sensor-based science
- Pervasive / ubiquitous computing,
human/animal behaviour analysis,
climate science, …
Some well known issues:
- Sensor reading quality – QA, outliers, false readings
- What we have: Metadata / context
- About the sensors id, type, calibration, parameter settings
- About the data readings timestamp
- About the quality assessed through QA processes
12. Preparedfor:
SystemsResearchChallengesintheInternetofThings
Newcastle,Jan.2016
IoT ∩ Science Metadata
This requires capturing and managing
provenance and other metadata
Provenance: a record of data derivation through multiple process
transformations
- Complementary to descriptive metadata
- enables reasoning about the findings, validation
• How was the data collected?
• How was it processed?
• Who was responsible?
16. Preparedfor:
SystemsResearchChallengesintheInternetofThings
Newcastle,Jan.2016
IoT ∩ Science
Typical uses for provenance:
• impact analysis (forwards)
• cause analysis (backwards)
Note on reproducibility: Observational data is generally not reproducible!
How much provenance is needed?
Impact analysis:
Suppose a sensor is later determined to be faulty (false readings)
How does that impact the experimental findings?
Cause analysis:
These conclusions seem implausible. What went wrong along the process?
17. Preparedfor:
SystemsResearchChallengesintheInternetofThings
Newcastle,Jan.2016
IoT ∩ People Personal Data from Things (PDT)
IoT vision: devices (“smart washing machine”) will make our lives better
They often also produce data that is also personal
As per the Data Protection Act 1998
• Are people aware of the trade-offs between privacy and benefits?
1. Ownership:
• What is “my” data? (who owns the utility consumption figures in my
house? Or an activity trace collected using a “smart shoe”?)
• Who else has access to it? To what extent?
2. Awareness of third party use of personal data:
• Who has been doing what with my data?
• How much of the data used in a certain computation is my data??
• What has its contribution been to the analytics?
3. Control. How much control can I have on the data that devices
produce on my behalf?
Ownership + awareness + control Trust
18. Preparedfor:
SystemsResearchChallengesintheInternetofThings
Newcastle,Jan.2016
Two recent publications
Mashhadi, Afra, Fahim Kawsar, and Utku Gunay Acer. “Human Data Interaction in IoT: The
Ownership Aspect.” In Internet of Things (WF-IoT), 2014 IEEE World Forum on, 159–162,
2014.
Vescovi, Michele, Corrado Moiso, Fabrizio Antonelli, Mattia Pasolli, and Christos
Perentis. “Building an Eco-System of Trusted Services through User Transparency,
Control and Awareness on Personal Data Privacy.” In Procs. W3C Workshop on Privacy
and User–Centric Controls. Berlin, Germany, 2014.
19. Preparedfor:
SystemsResearchChallengesintheInternetofThings
Newcastle,Jan.2016
IoT ∩ People Personal Data from Things (PDT)
Example:
SPHERE - a Sensor Platform for HEalthcare in a Residential Environment
(EPSRC, 2013-2018, Bristol, Prof. Ian Craddock)
http://irc-sphere.ac.uk/
Zhu. N, Diethe. T, Camplani. M, Tao. L, Burrows. A, Twomey. N, Kaleshi. D, Mirmehdi. M, Flach. P, Craddock. I, Bridging
eHealth and the Internet of Things: The SPHERE Project. IEEE Intelligent Systems 30 (4), 39-46. (doi: 10.1109/MIS.2015.57)
All about sensing, wearables, & detecting people’s activities
Instrumented “SPHERE house”
— scaling up to 100 homes by 2017 lots of data collection, data mining challenges
26. Preparedfor:
SystemsResearchChallengesintheInternetofThings
Newcastle,Jan.2016
Things that make decisions
Some challenges:
Provenance patterns for streaming, message passing: “V1 sent sij to V2”
How much “provenance” does each sensor reading need to carry? How
does this fit with M2M protocols?
Provlets: embed in messages vs stored separately in a repository
(indexed by key: <S.id, t>)
- M2M means more in-network provenance
- The data remains at the edge of the network
ETSI - European Telecommunications Standards Institute
Funding 11M
Types of sensors
SPHERE envisages sensors, for example:
1) That employ video and motion analytics to predict falls and detect strokes so that help may be summoned.
2) That uses video sensing to analyse eating behaviour, including whether people are taking their prescribed medication.
3) That uses video to detect periods of depression or anxiety and intervene using a computer-based therapy.
The SPHERE IRC will take a interdisciplinary approach to developing these sensor technologies, in order that:
1) They are acceptable in people's homes (this will be achieved by forming User Groups to assist in the technology design process, as well as experts in Ethics and User-Involvement who will explore issues of privacy and digital inclusion).
2) They solve real healthcare problems in a cost-effective way (this will be achieved by working with leading clinicians in Heart Surgery, Orthopaedics, Stroke and Parkinson's Disease, and recognised authorities on Depression and Obesity).
3) The IRC generates knowledge that will change clinical practice (this will be achieved by focusing on real-world technologies that can be shown working in a large number of local homes during the life of the project).
Configuration Storage Layer Contains “Configuration Storage API”.
The smart devices directly connect to this API during the bootstrap phase
It extracts the resource descriptions from the devices or (proxies in case of
legacy devices).
The layer houses a database and stores the device, endpoint and configuration resources in separate tables.
The API translates the CoRE Link based descriptions to appropriate storage format. This layer also keeps track of the configuration “lifetime” attribute.
During that period, if it does not receive an announcement that the device is still present or configuration update, it will delete that device configuration.
Sensor measurement alone has no value –
Need additional side information like unit, timestamp, type of sensor