Context Middleware for Adaptive Mobile Services

1
Context Middleware for Adaptive Mobile Services

• AWSI Workshop, June 2-3, 2005
• Dr. Theo G. Kanter,
• Ericsson and Wireless_at_KTH

2
Motivations

• Access to services via heterogeneous and multiple
  (operators) wireless networks.
• Interaction with multiple services using multiple
  devices.
• Services can make opportunistic use of available
  communication.

3
Adaptive Context-Aware Services

• Can no longer rely on a single network to arrange
  and manage these services.
• Adaptive services centered around the User
• Investigate new architectural choices that move
  service control and coordination closer to the
  user.
• Utilize context information
• User and / or communication
• Leverage information from sensors

4
Context Information NetworksService
Architecture Framework
5
CME
Application
Application Context API
CME
Context Socket API
Context Socket
Context Socket
(Remote) Context Sensing Context Data eXchange
Protocol (CDXP) Sensor Sampling Control Protocol
(S2CP)
MH
CH
6
Context Description Language

• Example of a context element
• ltacascontextelement id"123c"gt
• ltacasvalue datatype"integer"
• unit"temperature/kelvin"gt292lt/acasvaluegt
• ltacasentity-reference rel"acasdsv.su.se/k2/r
  7741/t"/gt
• ltacastimegtSat Apr 24 000521 CEST
  2004lt/acastimegt
• ltacassource uri "uriacasdsv.su.se/k2/csf/ap
  ax"/gt
• lt/acascontextelementgt

7
Peer Discovery in LAN / MAN

• LAN/MAN
• Service Peer Discovery Protocol (SPDP) uses
  SIP/SIMPLE and unicast (advantage wireless)
• CME ? SDPD-UA co-located with SIP-UA
• Better targeted at supporting service discovery
  in LAN/WAN than previous protocols, e.g. SLP.

8
Manage Protect Context Information Protecting
Privacy
Public Service Infrastructure
Anonymizer Proxy
PCM
CME
Context Generator (Hardware Sensors)
Context Generator (Software Sensors)
US
US
S
S
S
S
US
US
..
..
9
Context-Aware Mobile Services

• Context-Aware SIP Call/Invitation Delivery
• ? Allocated automatically with low delays to
  appropriate available device
• Context-Aware Call/Invitation Diversion
• ? Frees user from manually configuring device
  in/out of meeting mode
• Context-Aware Mobile Multi-purpose Audio
• device with storage leverages information from
  different context sources (e.g., storage space,
  battery, access conditions, location, ) local
  playout of cached content vs. streaming Vacas
• ? 2x the number of packets but 1/10 the battery
  consumption
• Context-Aware Opportunistic Communication
• Per packet policy decisions Mola ? better use
  of multiple interfaces ? avoid unnecessary
  vertical handoffs
• Transport context between devices Wennlund
• Knowledge of purpose of the communications
  expected (future) access conditions nearby
  devices ? Mechanisms for policy driven allocation
  of services Jansson, Avgeropoulos

10
ACAS User Study Spring of 2005

• 27 KTH students loaned an HP iPAQ 5550 handhelds
• Phase 1 Familiarization with the iPAQs
• Questionnaire indicated communication (important
  to the majority)
• ? enhance this with context sensitivity.
• Deployed Spot system similar other WiFi-based
  localization systems, except that localization
  provided by the system is user-relative.
• Phase 2 Users updated their positions as
  circumstances dictate
• Learned that this could be very infrequently
• Phase 3 Capitalize on users awareness
• studied impact on use of communication tools
• Combine user location with shortcuts to
  communication software
• Phase 4 Integrate ACAS components (undisrupted
  operation of prototype)
• Learned Cant rely on users to provide context
  (also causes privacy problems)
• ? Require automatic collection and processing as
  user utility hinges upon automatic and smooth
  operation

11
ACAS Contributions

• Relevant contributions in response to the
  challenges
• A service architecture, the context information
  network, to sense, manage, store, exchange of
  context information.
• Middleware components to prototype this context
  information network
• Context-aware mechanisms for service discovery
  service allocation and protecting user privacy
• Evaluation of location proximity sensing
• Early demonstrators of opportunistic wireless and
  mobile communication and prototypes, including a
  field trial

12
Conclusions

• 1. Adaptive Service DeliveryMobile-devices can
  become much smarter in communicating via
  policy-based decisions based on awareness of
  nearby devices, access networks, urgency ?
  Awareness enables more informed decisions and
  combined operation
• 2. Adaptive CompositionContext-awareness can
  free the user from manually configuring or
  selecting which device to use for which service.
• 3. Seamless Adaptive ServicesFurther optimize
  user utility of mobile media by clever
  (context-aware) use of available resources. ?
  automatic collection and processing. User utility
  hinges upon automatic and smooth operation.

13
Acknowledgements

• ACAS Project http//psi.verkstad.net/acas
• Research was conducted with partial support from
  the Swedish Foundation for Strategic Research
  (SSF), Ericsson, HP, Telia Sonera, Netlight, and
  R2M
• Contributors C-G Jansson, Martin Jonsson, Theo
  Kanter, Fredrik Kilander, Wei Li, Peter
  Lönnqvist, Gerald Q. Maguire Jr., Henrik
  Sidebäck, Tommy Westman, Andreas Wennlund, M.Sc.
  Students