A Knowledge-based Approach for Real-Time IoT Stream Annotation and Processing

1

Sefki Kolozali, Maria Bermundez, Daniel Puschmann, Frieder Ganz, Payam Barnaghi

Institute for Communication Systems (ICS)

University of Surrey

Guildford, United Kingdom

Smart Cities and Real-Time IoT Streams

− Data in smart cities is collected by sensor devices and also crowed sensing sources.

− The data is time and location dependent.− It can be noisy and the quality can vary. − It is continuous - streaming data

− Semantic annotation of data will help to describe:− provenance − spatial− temporal− thematic

Attributes of the data

3

The main objective

• to develop a framework in the scope of the CityPulse project for real-time IoT stream annotation that employs a knowledge-based approach to represent data streams and to support mashups.

• to develop an information model to represent abstract concepts and quality related attributes of IoT stream data.

• to enable delivery of large volume of data that can influence the performance of the smart city systems that use IoT data.

http://www.ict-citypulse.eu

4

The Key issues

• Virtualisation: Semantic annotation of heterogeneous data for automated discovery and knowledge-based processing• Heterogeneity

• Interoperability

• Aggregation and Abstraction: Large-scale data analytics• Data size

• Communication in distributed systems: exchange messages among different components• Time

• Space

• Synchronisation

5

Real-Time Stream Annotation Framework

6

Existing models - e.g. W3C SSN Ontology

Ontology Link: http://www.w3.org/2005/Incubator/ssn/ssnx/ssn

M. Compton et al, "The SSN Ontology of the W3C Semantic Sensor Network Incubator Group", Journal of Web Semantics, 2012.

Information Models

Describing a stream annotation work flow using the Stream Annotation Ontology (SAO)

Stream Annotation Ontology

.

.

.

The SAO allows representation of aggregated stream data and temporal characteristics. It is based on the SSN Ontology and Timeline Ontology.

IoT Stream Processing

WSNWSN

WSNWSN

WSNWSN

WSNWSN

WSNWSN

Network-enabled DevicesNetwork-enabled Devices

Network-enabled DevicesNetwork-enabled Devices

Network services/storage and processing

units

Data/service access at application level

Data collections and processing within the

networks

Query/access

to raw dataOr

Higher-level abstractions

MWMW

MWMW

MWMWData streamsData

streams

Middleware

Advance Message Queue Protocol (AMQP)

enum MType {transform,forward,store

}struct Message {

1: list<MType> messageTypes2: map<string,string> data3: map<string,string> metadata

}

• A publish/subscribe mechanism which decouples time, space and synchronisation.

• The message delivery logic lies with the message broker, decoupling it from the application layer.

Use Case Scenario- Traffic Scenario, Aarhus, DK

A visual representation of geographical coordinates on Google Map for a pair of road traffic sensors provided by city of Aarhus, Denmark.

12

Data abstraction

Using Symbolic Aggregate Approximation (SAX) and SensorSAX

SAX Pattern (blue) with word length of 20 and a vocabulary of 10 symbolsover the original sensor time-series data (green)

Source: P. Barnaghi, F. Ganz, C. Henson, A. Sheth, "Computing Perception from Sensor Data", in Proc. of the IEEE Sensors 2012, Oct. 2012.

fggfffhfffffgjhghfff

jfhiggfffhfffffgjhgi

fggfffhfffffgjhghfff

Data Aggregation with SAX and its representation based on SAO

@prefix sao: <http://example.com#> .@prefix ssn: <http://purl.oclc.org/NET/ssnx/ssn#> .@prefix qoi: <http://example.com/QoSQoI.owl#> .@prefix tl: <http://purl.org/NET/c4dm/timeline.owl#> .

:government a foaf:Organisation, prov: Agent .:sefki a foaf:Person, prov:Agent ;

foaf:givenName "Sefki" ;foaf:mbox <mailto:s.kolozali@surrey.ac.uk>prov:actedonBehalfOf :ccsrSurrey ; .

:sensorRec1 a sao:StreamData, ssn:SensorObservation ;prov: wasAttributedTo :government .

:sensorRec2 a sao:StreamData, ssn:SensorObservation ;prov: wasAttributedTo :government .

:traffic-sensor-recording-619 a sao:StreamEvent ; prov:used [ a sensorRec1; sensorRec2] ; sao:time [a tl:Interval; tl:at "2014-02-13T08:25:00"^^xsd:dateTime; tl:duration "PT15H30M"^^xsd:duration; ] ; prov:wasAsscoatedWith :sefki ; .:freshness-traffic-619 a qoi:Freshness ;

qoi:value "2014-02-13T08:25:00"^^xsd:dateTime .:sax_AverageSpeedSample a SymbolicAggregateApproximation;

rdfs:label "The sax representation of the traffic sensor recording obtained from Aarhus City.";sao:value "bbbbacdd";sao:alphabetsize "4"^^xsd:int ;sao:segmentsize "8"^^xsd:int ;prov:wasGeneratedBy traffic-sensor-recording-619; qoi:hasQoI freshness-traffic-619 .A real time average speed data obtained

from a pair of sensor points is mapped into SAX word, ”bbbbacdd”, with the segment size of “8” and alphabet size of “4” for 176 samples.

A excerpt from an RDF data annotated for a set of sensor recordings based on Stream Annotation Ontology.

Evaluation Results

In Conclusion

− We have developed a semantic model for the data streams in a smart city framework.

− The main advantages are providing an interoperable and machine-interpretable format for exchanging the data.

− The model can describe thematic, spatial, and temporal attributes of the streams and also the provenance data.

− It uses concepts from SSNO and ProvO. − We have also developed a message broker, wrapper (for restful services)

and a middleware to represent the data. − We also integrated it with a data abstraction method that we had developed

in our previous work.

− Future work: − We need to integrate this work with higher-level query mechanisms;− To integrate with our IoT data discovery and selection method;− Evaluate large-scale annotated data stream and query/access efficiency;

15

Q&A

− Thank you.

− EU FP7 CityPulse Project:

http://www.ict-citypulse.eu/

@ictcitypulse

p.barnaghi@surrey.ac.uk