Identification of malicious nodes in an AODV pure ad hoc network through guard nodes

This paper presents a guard node based scheme to identify malicious nodes in Ad hoc On-Demand Distance Vector (AODV) protocol. In this scheme each node calculates trust level of its neighboring nodes for route selection. Trust calculation process involves opinions of other nodes about the node whose trust level is to be determined. If a neighboring node has a trust level lower than a predefined threshold value, it is identified as malicious and it is not considered for route selection. The proposed model does not use any key distribution process and no changes are made in control packets of AODV. Simulation results show that the proposed scheme improves performance of AODV by identifying and removing malicious nodes. Performance of the scheme has been evaluated for three different types of malicious attacks (impersonation attack, colluding nodes attack and black hole attack).