Context-aware Infrastructure Issues

Nick TaylorHeriot-Watt University

Edinburgh, UK

DAIDALOS

What is Context-awareness?

• A range of information about the environment or state in which some activity is occurring

• For a network, context-aware also means being traffic-aware and for wireless, interference-aware

• Requirements for -– Services/Routes and SLAs (Service Level Agreements)– Bandwidth/Reliability and QoS (Quality of Service)– Frequency/Modulation and BERs (Bit Error Rates)

What Context-awareness Can Achieve

• Adaptation– Personalisation for users– Reconfiguration for devices

• Actuation– Pro-active behaviours for users– Switching for devices

• Monitoring– QoS for SLAs, etc. – Learning

Requirements of Context-awareness

• Rapid and reliable transmission of an increasing volume of, possibly continuously streamed, data

• Merging/fusion of data from multiple context sources

• Inference of higher level context attributes• For a network, cognitive network management

and re-configurability• Addressability of an enormous number of

sensors and actuators!

IPv4 Address Exhaustion

Source: Wikipedia (2010)

IPv6

• Slow uptake– Do we need to wait?– Can we afford to?

• Do we really need to be able to address everything from everywhere?– How far does RFID data actually travel in practice?– We need pragmatic solutions, not once-size-fits-all

• Ad hoc formation of local subnets on demand might be more efficient in most smart spaces– Opportunistic P2P networking is already possible but

needs to be made much easier

A Pervasive Agenda for FI

• Future Internet needs to better support– Smart spaces rich in devices• Including much larger numbers of, but highly localised,

sensors and actuators

– Mobile users on the move (across networks too)• B3G, WIFI, WIMAX, …• Separation of identity and location in addressing

– User choice and negotiation• Privacy, Cost, QoS, etc. selection criteria• Only the user can decide what their privacy, price, etc.

requirements are but negotiation must be supported

A Smarter FI Infrastructure

• Statefull– Awareness of what is occurring where (recoverable)

• Adaptable– Network selection and re-configuration

• Extensible– Integration of sensor and actuator networks

• Efficient– Network protocols (e.g. minimalist for mobiles)

• Seamless– Network hand-over for “Always Best Connected”

Users versus Devices

• Human to Machine (H2M)– Original Internet design– Needs extending to allow users to tailor more than just

apps/services to their needs• Cross-layer collaboration in Service Oriented Architectures• Preferences for cost, security, privacy, QOS need to be able to

impact on device and network properties

• Machine to Machine (M2M)– Data-centric rather than user-centric in nature– Real World Internet and Internet of Things will drive FI in

new directions with potential for conflict with H2M unless infrastructure sufficiently flexible

Acknowledgements

• Partners in Daidalos, Persist, Societies consortia

• MANA (Management & Service-Aware Networking Architectures) Position Paper

• RWI (Real World Internet) Position Paper

• Additional slides follow in case necessary

PERSIST (PSS) & SOCIETIES (CSS)

Example - PERSIST

Example - SOCIETIES

• Mobile Devices: SmartPhone, MID, Tablet PC, PDA, Nettop, Specks/Motes, RFID Tags/Readers

• Fixed Devices: Sensor networks, Servers, IMS, Directional Speakers, Lights

• Available Features: GPS, Motion, Temperature, Pressure, Magnetic Compass, Camera, Microphone

• OS Platforms: Windows, Linux, Android, Moblin

• Networking: LAN, Wifi, WiMax, UMTS, BGAN, Bluetooth, Zigbee

• Service Interaction: REST, SOAP, PTP (Cameras), Sensor Network Protocols

• Service Platforms: OSGi, Mono

•User Trial/Testbed Technologies

•Abstraction Technologies

• Protocols: IP, HTTP, FTP (Cameras)