A fault diagnosis method based on decision tree for wireless mesh network

Fault diagnosis for wireless mesh network is an active field in recent years, and also the decision tree algorithm is widely used in Data Mining field. How to apply machine learning algorithm in network fault diagnosis presents challenge. This paper proposes a rule post-pruning method named as W-C4.5-RP which is based on traditional C4.5 algorithm. In order to verify the validation of the algorithm, we trained the decision tree by injecting faults actively into wireless mesh network (WMN) in a testbed in campus networks. The experimental results show that the W-C4.5-RP algorithm can shorten diagnostic time and increase diagnostic efficiency. This method can be applicable for the faults diagnosis in wireless mesh networks which are characteristic of burst and extremely short duration.