Mobile Cloud and Crowd Computing, Communications and Sensing

Prof Ivan Stojmenovic (University of Ottawa, Canada) will give a talk on “Mobile Cloud and Crowd Computing, Communications and Sensing”.

Abstract Mobile devices (smart phones, tablets, laptops, embedded boards, robots) can serve as terminals for cloud computing services over intelligent network. Mobile cloud has emerged as a new cloud computing platform that 'puts cloud into a pocket'. Important issues include optimizing the scheduling and transport schemes, access management, and application optimization, for mobile devices to achieve energy saving. This talk will first introduce the development of mobile cloud computing and describe some applications involving multimedia, vision/recognition, graphics, gaming, text processing. It will present the transmission, computation (e.g. task outsourcing), and sensing (e.g. location based services) challenges and solution approaches of green computing in mobile cloud. 'Crowd computing' combines mobile devices and social interactions to achieve large-scale distributed computation. Examples include task farming and social network creation and cooperation. Mobile devices are being equipped with various sensors to provide input for participatory and opportunistic crowd-sourced sensing. Particular emerging concepts are 'vehicular cloud' and 'vehicular crowd', with applications such as cloud server, vehicular data center, and congestion mitigation.

Biography http://www.site.uottawa.ca/%7Eivan/Stojmenovic-short-cv.pdf

• Program partition
  • kumar, 2011, mobile server
• Cloudlet cmu, where the latancy is high. data center in a box.
• Hyrax: Multimedia search and sharing app.

• Robot: peer, clone.
• Biometric apps: input from mobile devices, preprocessing( reduce data size), backend processing
• e.g.: image search, real time forensic.
• Socialize spontaneously with mobile app. Li infocom 2012
• Adding proxy cloud interaction: composer collaborate in realtime.
• eSmall talker: find potential small talk oppotunities.
• From Cloud to crowd computing. ←— this is it
  • remove cloud: computing in mobile
  • Spontaneous wireless ad-hoc network.
  • Authentication issues: AES
  • trust issues
• Cloud for language translation
• Social aware networking
  • VANs,
• Architecture of Social Area networking
  • sensing wiht mobile device
  • l2: learning
    • communitz, similaritz, centralitz, obility pattern
  • l3 network protocol
  • l4 application
• Crowd computing
  • Combining mobile devices and social interaction to achive large scale computation
    • Task farming
    • Social aware task farming
      • social netowrk formed by human interaction
      • community structured.. DTN??
  • People centric sensing
    • sharing sensing presence
    • crowd-sourced sensing and collaboration using twitter.
• Research challenges
  • localized analytic: context inference.
  • resource limit: energy, bandwidth, computation.
  • privacy, security
  • aggregate analytics: data mining
  • architecture: unify for different applications, cooperation in sensing

• smart vehicles
• ITS application system model
  • car to car, car to infrastructure (roadside unit)
  • collaboration congestion detection.
• Cloud service by a vehicle with RSU t oRSU service connection.
• Cloud in the parking lot.
• Datacenter at the shopping mall –> cars buy and sell electricity.
• Traffic jam resolution, evacuation.
• Network as a service, sharing network resource between cars. –> interesting

• uer side of green cloud
• Transmission
  • energy cost on mobile
  • unstable wireless quality
  • heterogeneous ifaces
  • different traffic demand
• computation
  • limited resources
  • finite energy source
  • task out-sourcing schedule
    • which can be offloaded
    • profile applications based on energy: energy state prediction, power modeling.
      • non utilization system calls slowly change power state
      • several components donot have quantitative utilization.
        • non utilization based power modeling.
      • energy saving of location based service
        • subtitution: cell based location tracking: sensor, gps, cell
        • suppression: suppress unnecessary invocation of location sensing when no bility, app stop
        • piggybacking: duplicate location sensing.
  • cpu optimization

• conclusion:
  • cloud computing will be extended to vehicular assets from individual vehicles to entire fleets,
  • cell phones and other commodity cnsumer products.

• Note: offloading methods
  • virtualiyation cloudlet, MAUI, cloneCloud,
• offloading: study wireless communication, link information to make decision on offloading.