RFIDbased Distributed Memory for Mobile Applications

1
RFID-based Distributed Memory for Mobile
Applications

• Michel Simatic

2
Table of contents

• Introduction
• Passive RFID tags in existing architectures
• RFID-based distributed memory
• Experimental results
• Issues
• Conclusion and future work

3
Table of contents

• Introduction
• Passive RFID tags in existing architectures
• RFID-based distributed memory
• Experimental results
• Issues
• Conclusion and future work

4
IntroductionRFID overview

• RFID tag Memory Antenna (radio)
  communication capabilities
• Memory contains
• (always) A unique identifier
• (possibly) Bytes which can be written (and read
  -) )
• Tags can be passive or active
• NFC technology can be used to interact with
  nearby tags
• By 2013, 700 million users will have an
  NFC-enabled mobilephone NFC study, 2008
• An RFID tag contains also applicative data An
  RFID tag stores applicative data (located in
  the tag or elsewhere) related to the physical
  entity (object, person, location) which the tag
  is linked to

5
IntroductionClassical RFID-based architectural
pattern

• Centralized architectural pattern GS1 EPCglobal,
  2007

6
IntroductionClassical RFID-based architectural
pattern
Write DEF
2 Request to write DEF on tag r1
1 Read tag Id
7
IntroductionClassical RFID-based architectural
pattern
Read tag
2 Request value of tag r2
1 Read tag Id
Request value of tag r2
Query tag r2
8
IntroductionClassical RFID-based architectural
pattern

• Problem with centralized architectural
  patternIt requires an access to a global
  network
• LAN (Local Area Network)The mobile device must
  be connected via a cable or Bluetooth to an
  entity connected to a LAN
• WLAN (Wireless Local Area Network)The mobile
  must have Wi-Fi capabilities The area must be
  covered by Wi-Fi
• WAN (Wireless Area Network)The mobile must have
  access to a data plan allowing GPRS, UMTS or
  HSDPA connections

9
IntroductionGoal of this study

• A global network is not available for some
  RFID-based applications
• Find an RFID-based architectural pattern which
  meets the following requirements
• Use of passive tags
• No use of a global network (LAN, WLAN or WAN)
• When users equipped with an RFID-enabled device
  read a tag, the applicative contents which is
  read is correct (no staleness issue)
• Users equipped with an RFID-enabled device can
  query about the applicative contents of any
  remote tag (no need to be physically near the tag)

10
Table of contents

• Introduction
• Passive RFID tags in existing architectures
• RFID-based distributed memory
• Experimental results
• Issues
• Conclusion and future work

11
Passive RFID tags in existing architectures
Centralized architectural pattern

• How does it meet the requirements?
• No use of a global network
• When users equipped with an RFID-enabled device
  read a tag, the applicative contents which is
  read is correct (no staleness issue)
• Users equipped with an RFID-enabled device can
  query about the applicative contents of any
  remote tag (no need to be physically near the tag)

12
Passive RFID tags in existing architectures
Semi-distributed architectural pattern

• Pattern used in the application used to take care
  of Paris trees Paris trees, 2006

13
Passive RFID tags in existing architectures
Semi-distributed architectural pattern
Sync
Sync
Sync
Sync
Sync
Sync
14
Passive RFID tags in existing architecturesSemi-d
istributed architectural pattern
Write DEF
2 Write DEF
1 Read tag Id
15
Passive RFID tags in existing architecturesSemi-d
istributed architectural pattern
Read tag
2 Read contents
1 Read tag Id
Read contents
Query tag r2
16
Passive RFID tags in existing architecturesSemi-d
istributed architectural pattern
1 Read tag Id
Read tag
Read contents
??????
Read tag
1 Read tag Id
Read contents
17
Passive RFID tags in existing architecturesSemi-d
istributed architectural pattern
Sync
Sync
Sync
Sync
Sync
Sync
18
Passive RFID tags in existing architecturesSemi-d
istributed architectural pattern

• How does it meet the requirements?
• No use of a global network
• When users equipped with an RFID-enabled device
  read a tag, the applicative contents which is
  read is correct (no staleness issue)
• Users equipped with an RFID-enabled device can
  query about the applicative contents of any
  remote tag (no need to be physically near the tag)

19
Passive RFID tags in existing architecturesDistri
buted architectural pattern

• The Ubiquitous Near-Field Distributed Memory
  Couderc and Banâtre, 2009

20
Passive RFID tags in existing architecturesDistri
buted architectural pattern
Write DEF
Write DEF
21
Passive RFID tags in existing architecturesDistri
buted architectural pattern
Read tag
Read contents
???????????
Query tag r2
22
Passive RFID tags in existing architecturesDistri
buted architectural pattern

• How does it meet the requirements?
• No use of a global network
• When users equipped with an RFID-enabled device
  read a tag, the applicative contents which is
  read is correct (no staleness issue)
• Users equipped with an RFID-enabled device can
  query about the applicative contents of any
  remote tag (no need to be physically near the tag)

23
Table of contents

• Introduction
• Passive RFID tags in existing architectures
• RFID-based distributed memory
• Experimental results
• Issues
• Conclusion and future work

24
RFID-based Distributed Memory

• Mixing Semi-Distributed and Distributed
  architectural patterns

25
RFID-based Distributed Memory
Write DEF
1 Read contents
2 Write DEF and 42 (411)
26
RFID-based Distributed Memory
Read tag
1 Read contents
2 Write contents
Query tag r2
27
RFID-based Distributed Memory
1 Read
2 Write
Read tag
Read tag
1 Read
2 Write
28
RFID-based Distributed Memory

• How does it meet the requirements?
• No use of a global network
• When users equipped with an RFID-enabled device
  read a tag, the applicative contents which is
  read is correct (no staleness issue)
• Users equipped with an RFID-enabled device can
  query about the applicative contents of any
  remote tag (no need to be physically near the tag)

29
Table of contents

• Introduction
• Passive RFID tags in existing architectures
• RFID-based distributed memory
• Experimental results
• Issues
• Conclusion and future work

30
Experimental resultsPervasive game as an
application example

•  Plug Secrets of the museum  a pervasive
  game played inMusée des arts et métiers Plug,
  2009
• 12 game sessions / 150 players15 game sessions /
  200 players
• 16 RFID tags are linked to specific objects. Each
  tag hosts a virtual card
• Players are equipped with NFC-enabled mobile
  phones. Each phone hosts4 virtual cards
• Players have to collect virtual cards of the same
  family.To do so, players exchange virtual cards
  with tags or with other players
• No global network
• Nokia 6131 NFC is not Wi-Fi enabled ( the museum
  is not covered with Wi-Fi)
• UMTS data plans are too expensive
• The distributed architectural pattern could fit,
  but

31
Experimental resultsPervasive game as an
application example

• but the game design requires a hint function
• To help players collecting their family of cards,
  players can ask their mobile for a hintTwo
  possible answers
• 2 minutes ago, card Lavoisier was on tag
  Statue of Liberty
• A priori, no tag contains an object which is of
  interest for you
• To compute this answer, the mobile must query the
  contents of all of the tags, including the remote
  ones.? RFID-based Distributed Memory
• DMer is a byte containing the value of the
  virtual card hold by tag rVCer is a short
  containing the vector clock value

32
Experimental results

• 5 sessions (with 6 players each) were analyzed to
  evaluate RFID-based Distributed Memory
• 88 of the hints were correct
• Correctness depends of
• How many tags are notified of a change
• How long it takes to disseminate such information

33
Experimental results

• To evaluate dissemination of stale information,
  we define the notion of validity period
• Validity period Time of the next change of
  tag r Time at which information concerning
  tag r contents is written

Tag r3
Write MNO Tag r1 contains DEF Tag r2 contains UVW
Tag r2
Write RST Tag r1 contains GHI
Write UVW Tag r1 contains DEF
Tag r1
Write DEF
Write GHI
Blue mobile
Orange mobile
Time
Validity period of r1 info t3 t4 lt 0 Stale
information
Validity period of r2 info t5 t4 gt
0 Up-to-date information
t1
t2
t3
t4
t5
34
Experimental results

• Frequency of validity periods

)
35
Table of contents

• Introduction
• Passive RFID tags in existing architectures
• RFID-based distributed memory
• Experimental results
• Issues
• Conclusion and future work

36
Issues

• Staleness
• Experimental results
• 100 88 12 of hint are not correct
• 5 of data remained stale more than 12 minutes
• How to reduce the problem?
• A dedicated user goes periodically through all of
  the tags
• We ask all users to periodically meet all
  together to synchronize their DMmobile
• We relax the constraints concerning the global
  network.Thus, we introduce a server.Periodically
  , each phone synchronizes DMmobile and DMserver
• Scalability
• Hardware has limited size
• It takes time to read DMr and VCr, and to write
  them back
• For Plug Secrets of the museum, limit is
  min(341,1104) 341 tags

37
Table of contents

• Introduction
• Passive RFID tags in existing architectures
• RFID-based distributed memory
• Experimental results
• Issues
• Conclusion and future work

38
Conclusion

• RFID-based Distributed Memory meets the
  requirements
• No use of a global network (LAN, WLAN, WAN)
• When users equipped with an RFID-enabled device
  read a tag, the applicative contents which is
  read is correct (no staleness issue)
• Users equipped with an RFID-enabled device can
  query about the applicative contents of any
  remote tag (no need to be physically near the
  tag)
• It is an interesting alternative to existing
  architectural patterns
• Integrate in DM data linked to other entities of
  thesystem (e.g. mobile devices). See demo.
• RFID-based Distributed Memory can be considered
  as an implementation of opportunistic data
  flooding Zebranet, 2002.Our contribution is to
  integrate passive RFID tags in such peer-to-peer
  architecture

39
Future work

• Tackle issues
• Staleness
• Scalability
• RFID-based Distributed Memory works because three
  conditions are verified
• An element of DM can only be modified in one
  place
• The set of RFID tags contributing to DM is
  defined at the beginning of the system lifetime
• The set of RFID tags is ordered
• Release one (or more) of the three conditions
• Model this Distributed Memory to determine the
  conditions on the number of mobiles, the behavior
  of the mobile users, the number of tags, the
  distance between the tags which turns this
  Distributed Memory into a useful architecture

40
Thank you for your attention

• Questions ?

41
References

• Couderc and Banâtre, 2009 P. Couderc and M.
  Banâtre. Beyond RFID The Ubiquitous Near-Field
  Distributed Memory. ERCIM news, (76)3536,
  January 2009.
• GS1 EPCglobal, 2007 F. Armenio, H. Barthel, L.
  Burstein, P. Dietrich, J. Duker, J. Garrett, B.
  Hogan, O. Ryaboy, S. Sarma, J. Schmidt, K. Suen,
  K. Traub, and J. Williams. The EPCglobal
  architecture framework. Technical Report Version
  1.2, GS1 EPCglobal, September 2007.
• NFC study, 2008 Christian D. 700 million of
  users of NFC mobiles in 5 years (in French).
  http//www.generation-nt.com/juniper-etude-technol
  ogie-nfc-mobile-utilisateurs-actualite-151831.html
  , September 2008.
• Paris trees, 2006 ITR Manager.com. City of
  Paris is taking care of its trees with RFID tags
  (in French). http//www.itrmanager.com/articles/59
  758/59758.html, December 2006.
• Plug, 2009 M. Simatic, I. Astic, C. Aunis, A.
  Gentes, A. Guyot-Mbodji, C. Jutant, and E. Zaza.
  Plug Secrets of the Museum A pervasive game
  taking place in a museum. In Entertainment
  Computing - ICEC 2009, Eighth International
  Conference, Paris, France, September 3-5, 2009,
  Proceedings, Lecture Notes in Computer Science.
  Springer, September 2009.
• Zebranet, 2002 P. Juang, H. Oki, Y. Wang, M.
  Martonosi, L. S. Peh, and D. Rubenstein.
  Energy-efficient computing for wildlife tracking
  design tradeoffs and early experiences with
  zebranet. In ASPLOS-X Proceedings of the 10th
  international conference on Architectural support
  for programming languages and operating systems,
  pages 96107, New York, NY, USA, 2002. ACM.