See also: IRC log
<trackbot> Date: 24 June 2015
<scribe> meeting: Thing Descriptions
The agenda:
1) Preliminaries
2) Presentation: Standardized and Efficient RDF Encoding (Sebastian Käbisch, Siemens)
3) Plans for use case and requirements discussions
4) Discussion/Brainstorming
<scribe> scribenick: dsr
Sebastian notes that there will be a joint call with the security task force on July 8.
Sebastien presents some slides (URI to follow in email)
The challenge is to understand what capabilities/features a given IoT node has.
Interest in using standardised semantic web technologies as a way to address this (RDF, OWL, SPARQL)
There are several existing serialisation formats for RDF, e.g. N-triples RDF/XML, and Turtle. However these can be difficult to support on constrained devices.
Typical IoT sensors have limited functionality and limited network bandwidth.
Data values are typically numbers not strings in order to represent physical values.
There are several approaches to compressing data (slide5 with table)
ZIP isn’t a good fit. CBOR is better, and EXI even better, but at the cost of requirement of prior agreement on the data model.
EXI is a W3C standard, and ongoing work is making it less dependent on XML
Sebastian presents an RDF/XML example for a temperature and humidity sensor
EXI applies the RDF EXI grammar to create an efficient binary representation of the sensor data
EXI doesn’t require a grammar, but given one can produce better compression.
Slide 10 presents a micro RDF store together with REST messages over CoAP
Slide 11 presents data on compression ratio versus the size of the data set for various different flavours of EXI encoding.
Slide 12 gives data on memory usage for an ARM Cortex M3 chip with 64KB RAM and 256KB ROM
Conclusion: we can combine RDF based semantics with microcontrollers when using EXI
Questions?
Dave: I don’t think we can
mandate a single data encoding format, although we can offer
advice for helping people to choose for themselves
... secondly, the thing description models which are based upon
Linked Data should support efficient encodings
Dave notes that Carsten Bormann is a strong supporter of CBOR and perhaps we should invite his comments.
Sebastian: strings can be a problem for some encodings
Danh: it isn’t clear what you’re using EXI for?
Sebastian: it is mainly for serialisation of data
Moreover, we can store the EXI in a database for efficient storage
Danh: did you do a comparison with binary formats for JSON
Sebastian: EXI tends to do better than some JSON binary encodings due to better handling of strings
Daniel: Qualified names disappear in JSON
CBOR is better than some JSON binary encodings.
Danh: due to MCU resource constraints, regular compression is impractical, right?
Sebastian leaves for a train …
Daniel: for ZIP you need a large buffer while the dictionary is being constructed. EXI avoids that through prior compilation of the grammar
[meeting ends abruptly for scribe]
<DanhLePhuoc> webex stopped for me
This is scribe.perl Revision: 1.140 of Date: 2014-11-06 18:16:30 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/Anicicm/Anicic,/ Succeeded: s/prior/of prior/ Found ScribeNick: dsr Inferring Scribes: dsr Present: Dave_Raggett Sebastian_Kaebisch Arne Ashok_Malhotra Danh_Le_Phuoc Daniel_Peintner Darko_Anicic Hamid_reza_Laleh Joel Found Date: 24 Jun 2015 Guessing minutes URL: http://www.w3.org/2015/06/24-wot-td-minutes.html People with action items: WARNING: Input appears to use implicit continuation lines. You may need the "-implicitContinuations" option.[End of scribe.perl diagnostic output]