The Application Of Reputation Model ForWMN Intrusion Detection

Wireless mesh network (WMN) is a wireless broadband network based on IP Protocol. As an integration of the advantages of WLAN and Ad Hoc network, WMN is characterized of quick self-organization, high spectral efficiency, and great scalability. Its’Multi-hop routing mechanisms enhance network coverage and flexibility, while increase the difficulty of ensuring the security of all the nodes. Due to the wireless network infrastructure and the mobility of communication nodes, it is difficult to control the access of users and nodes through boundaries protective equipments such as firewall and proxy server, thus WMN can be vulnerable to the attack of malicious nodes. So, to ensure the security of the network, an instrusion detection mechanism aiming at malicious nodes should be used in the application of WMN.The reputation mechanism evaluates the security status of the nodes mainly by evaluating their behavior. This paper introduces the reputation model into WMN’s intrusion detection system, which dectects the anomalous nodes by judging the reputation ones. Unlike Ad Hoc network and P2P network, WMN are generally huge in scale, which makes the convergences of ordinary prestige models slow and resource-consuming. Thus this paper proposes lamination reputation management model, in which WMN are clustered on the base of the ordinary reputation model. Cluster heads are considered as management nodes, and the reputation is formed and transported in the clusters with a narrowed transmission scope of reputation, increased reputation convergence speed, and decreased network resources consumption. The concept of trust intensity is proposed to distinguish and choose the right trust recommendation for different links. An anomalous nodes detection algorithm based on reputation track is proposed in this paper, which differs from traditional strategy that judges the anomalous nodes just by the changes of reputation values of their own nodes. This algorithm can make the reputation-based intrusion detection more precise and then reduce the rate of false positives.In the aspect of model structure, the management node disaster recovery mechanism is proposed to solve the problem of excessive reliance on management nodes in reputation centralized management system. Under this mechanism, even when management nodes are under attack, the reputation system can still run smoothly, which reduces the limitation of choosing nodes as management nodes.The simulation results show that the model can quickly and accurately indenify the anomalous nodes, ensure the safe and sound operation of WMN even under invasion, and achieve an effective utilization of network resources.