摘要： 针对传统的分簇方法很少考虑安全因素，或者只考虑安全性而忽视对网络性能影响的问题，提出一种基于信任关系的分簇方法。该方法结合人类记忆的扩散激发模型的思想，能够根据有限的局部信息，自动地对整个网络进行分割，在提高Ad Hoc网络性能的同时，还可提高其安全性。实验结果表明，该分簇方法在精确度方面与集中式的分簇方法非常接近。

关键词: 簇, 分簇, 信任, 扩散激发

Abstract: Few security factors are taken into account in clustering traditionally and impact on network performance is always ignored. This paper proposes a novel approach for trust relation based clustering in Ad Hoc network. The approach is based on spread activation model, which is an organization structure for human long-term memory. This scheme can partition the whole network into many clusters according to limited local knowledge automatically. Simulation results show the accuracy of the clustering approach is very closed to the concentrating method.

Key words: cluster, clustering, trust, spread activation

中图分类号: 

• TP391