A Highly Adaptable Reputation System Based On Dirichlet Distribution

Conventional reputation system follows 2-dimensional Beta-distribution to do one rule observation. However, complex environments with high BER and propagation delay, strong interference or moving protocol will cut down the accuracy of detection. Apart from that, the response times of reputation value are various from different circumstances, which demand an adaptable mechanic. Therefore in this paper, we propose a highly adaptable reputation system with the ability of results category, fluctuation toleration, response time adjustment, and attack prevention. The results show our system has great performance on fuzzy state detection, effectively filters the burst fluctuation and rapidly responds to the long term variation with low power cost. Practically, the performance under wormhole attacks such as jam-and-reply attacks is promising.