Subscribe: Contact: Sitemap: Printer Friendly
Subscribe to our RSS feed
Home  >  People  >  Sebastien Ardon  >  ubistore

ubistore

ubistore




Keywords: opportunistic forwarding, storage

Participants: Sebastien Ardon, Feiselia Tan, Max Ott, Youmna Borghol

Context

Digital Media is pervasive to our lives: digital photographs, music, communications. Users are carrying portable and embedded devices: digital cameras, portable audio devices, laptop, PDA and next devices These devices, before used mainly for communicating, are now used as storage devices. Their capacity is increasing exponentially (e.g. 80GB ipod)

Data Backup is seen as preventive medicine: “a device failure won’t happen to me” Existing backup systems are centralized systems: as reliable as your backup server/hard disk/CDROM. Most systems are cumbersome and require manual user intervention. 

What is needed is an automatic and ideally operating cost free backup solution for mobile devices. While many of these devices have, or soon, will have wireless data connectivity, the data volume and associated network cost makes traditional backup to a backup server cost-prohibitive for most consumers. Finally, the 100% reliability requirement of traditional backup systems is arguably unecessary for many applications where a partial recovery of a user's "digital life" would be acceptable given the free nature of the service.

The UbiStore approach

UbiStore is a novel distributed and opportunistic backup architecture, where mobile devices backup their data over shortrange, ad-hoc wireless links to other devices encountered as a result of user mobility. We therefore propose to extend the concept of social backup to a mobile environment, where mobile devices share stor- age space to backup data. 

We base our design on the assumption that typical user mobility patterns will incur some repetitive encounters in the course of daily life (public transports, home/office), which are sustained on large time scales and can therefore facilitate the recovery of data in case of a device failure. We set out to design a data backup architecture which exploits both this repetitiveness and diversity of human inter-contacts on the day to week timescale.


Challenges

  • Communication: swarming, asynchronous, intermittent connectivity. Delay-Tolerant Network and other opportunistic communication application
  • Mobility Models
  • Security and trust models
  • Privacy criterias
  • Load balancing and fairness
  • Consistency, data encoding (erasure codes, etc.)
  • Evaluation methodology
    • define reliability metric
    • define backup recovery speed metric

Status

  • First generation storage algorithm
    • Neighbour selection
    • Communication protocol over Bluetooth technology
    • Block management strategy
  • Prototyping and simulation over JIST+SWANS mobile ad-hoc simulator
  • Scalable mobility Modelling using EMO:
    • encounter trace statistical analysis
    • scalable simulation of key encounter parameters
    • integration within JIST

 

Publications

  • F. Tan, S. Ardon, M. Ott. UbiStore: Ubiquitous and Opportunistic Backup Architecture, IEEE ICMAN workshop. In Proc. Of Fifth Annual IEEE Conference on Pervasive Computing and Communications, ICMAN Workshop (PerCom 2007), White Plains, NY, USA, March 2007
  • F. Tan, S. Ardon, R. Hsieh. The Impact of User Mobility Patterns on Opportunistic Content Distribution Network. To appear in Proc. of IEEE International Conference on Networks (ICON 2007), Adelaide, Australia, October 2007.
  • F. Tan, S. Ardon, Y. Borghol. EMO: A Statistical Encounter-based Mobility MOdel for Simulating Delay Tolerant Networks. Submitted to the 9th IEEE International Symposium on a World of Wireless, Mobile and Multimedia Networks (WoWMoM 2008)