Detecting packet-dropping faults in mobile ad-hoc networks

Mobile ad-hoc networks are inherently prone to security attacks, with node mobility being the primary cause in allowing security breaches. This makes the network susceptible to Byzantyne faults with packets getting misrouted, corrupted or dropped. In this paper we propose solutions using an unobtrusive monitoring technique using the "detection manager" to locate malicious or faulty nodes that misroute, corrupt or drop packets. The unobtrusive monitoring technique is similar to an intrusion detection system that monitors system activity logs to determine if the system is under attack. This technique uses information from different network layers to detect malicious nodes. The detection manager we are developing for mobile ad-hoc networks stores several rules for responding to different situations. Any single node in the network can use unobtrusive monitoring without relying on the cooperation of other nodes, which makes unobtrusive monitoring easy to implement and deploy. Simulations of mobile ad-hoc networks that contain malicious nodes indicate that unobtrusive monitoring has a high detection effectiveness with low false positive rate.