2nd International ICST Conference on Mobile and Ubiquitous Systems: Networking and Services

Research Article

A reputation-based mechanism for isolating selfish nodes in ad hoc networks

  • @INPROCEEDINGS{10.1109/MOBIQUITOUS.2005.7,
        author={M.T.  Refaei and Vivek  Srivastava and L. DaSilva, and M. Eltoweissy},
        title={A reputation-based mechanism for isolating selfish nodes in ad hoc networks},
        proceedings={2nd International ICST Conference on Mobile and Ubiquitous Systems: Networking and Services},
        publisher={IEEE},
        proceedings_a={MOBIQUITOUS},
        year={2005},
        month={11},
        keywords={},
        doi={10.1109/MOBIQUITOUS.2005.7}
    }
    
  • M.T. Refaei
    Vivek Srivastava
    L. DaSilva,
    M. Eltoweissy
    Year: 2005
    A reputation-based mechanism for isolating selfish nodes in ad hoc networks
    MOBIQUITOUS
    IEEE
    DOI: 10.1109/MOBIQUITOUS.2005.7
M.T. Refaei1, Vivek Srivastava1, L. DaSilva,1, M. Eltoweissy1
  • 1: Bradley Dept. of Electr. & Comput. Eng., Virginia Univ., Alexandria, VA, USA

Abstract

For ad hoc networks to realize their potential in commercial deployments, it is important that they incorporate adequate security measures. Selfish behavior of autonomous network nodes could greatly disrupt network operation. Such behavior should be discouraged, detected, and isolated. In this paper, we propose a reputation-based mechanism to detect and isolate selfish nodes in an ad hoc network. The proposed mechanism allows a node to autonomously evaluate the "reputation" of its neighbors based on the completion of the requested service. The underlying principle is that when a node forwards a packet through one of its neighbors, it holds that neighbor responsible for the correct delivery of the packet to the destination. Our mechanism is efficient and immune to node collusion since, unlike most contemporary mechanisms for reputation-based trust, it does not depend on exchanging reputation information among nodes. We also explore various reputation functions and report on their effectiveness in isolating selfish nodes and reducing false positives. Our simulation results demonstrate that the choice of the reputation function greatly impacts performance and that the proposed mechanism, with a carefully selected function, is successful in isolating selfish nodes while maintaining false positives at a reasonably low level.