A simple scheme for pseudo clustering algorithm for cross layer intrusion detection in MANET

The Mobile AdHoc Network (MANET) is a type of wireless network that does not require infrastructure for its operation; therefore, MANETs lack a centralized architecture which affects the level of security inside the network and increases vulnerability. Although encryption helps to increase network security level, it is not sufficient to protect against malicious intruders. An intrusion detection scheme is proposed in this paper based on cross layer feature collection from the medium access control (MAC) and network layers. The proposed method employs an hierarchical configuration that avoids using a clustering algorithm and, instead, sequentially activates the promiscuity (ability to sniff all packets transmitted by nodes within radio range) of the node based on its location in the network. The node in this case acts as a pseudo cluster head (PCH) that collects data from its neighboring nodes in each quadrant in the field and then uses this information to calculate an anomaly index (AI) in each quadrant. The mechanism uses a C4.5 decision tree to learn the network behavior under blackhole attack and is able to recognize blackhole attacks with up to 97% accuracy. The presented approach is twofold - it is energy efficient and has a high degree of intrusion detection with low overhead.