Simulation of a distributed recommendation system for pervasive networks

  • Authors:
  • Seamus Moloney

  • Affiliations:
  • Nokia Research Center, Helsinki, Finland

  • Venue:
  • Proceedings of the 2005 ACM symposium on Applied computing
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

New networks based on short-range radio are emerging where mobile devices interact in a spontaneous and short-lived fashion. Such pervasive networks involve devices approaching and finding each other, interacting to exchange some information or use some shared service, before disconnecting and leaving radio range. As such, they involve lots of potentially risky interactions with strangers. Due to the lack of any infrastructure in these networks, using a distributed recommendation system to improve security is an attractive option. This paper describes a simulation of a simple distributed recommendation system running in such a network. These networks can be modeled as social networks and the simulation builds on an existing algorithm that has been shown to model social networks accurately. The results of the simulation suggest that with a relatively small overhead, the trust information can move effectively in such networks and the system can deliver a substantial decrease in the level of risk these spontaneous interactions involve.