MODELING TRUST IN SELF-ORGANIZING SYSTEMS WITH HETEROGENEITY

摘要

Self-organizing systems (SOS) possess the potential of performing complex tasks in uncertain situations with adaptability. Despite the benefits of self-organizing, it is also subject to the influence of unpredictable behavior of individual agents and environment noises. In hostile situations for example, individual performance of self-organizing agents may deteriorate and this can lead to system malfunction, posing great challenge for the design of SOS. In this paper, we propose a trust based model as a design approach to SOS in consideration of the capability heterogeneity of the agents. A box-pushing task was presented and studied. Trust is measured using beta probability distribution, which takes into account both the positive and negative interactions between the agents. The simulation results have shown that our trust model ensures favorable interactions among agents and leads to increased system effectiveness and conditional system efficiency improvement in comparison to SOS without using a trust model.